You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

203 lines
5.4 KiB

cs: sync WAL more frequently (#3300) As per #3043, this adds a ticker to sync the WAL every 2s while the WAL is running. * Flush WAL every 2s This adds a ticker that flushes the WAL every 2s while the WAL is running. This is related to #3043. * Fix spelling * Increase timeout to 2mins for slower build environments * Make WAL sync interval configurable * Add TODO to replace testChan with more comprehensive testBus * Remove extraneous debug statement * Remove testChan in favour of using system time As per https://github.com/tendermint/tendermint/pull/3300#discussion_r255886586, this removes the `testChan` WAL member and replaces the approach with a system time-oriented one. In this new approach, we keep track of the system time at which each flush and periodic flush successfully occurred. The naming of the various functions is also updated here to be more consistent with "flushing" as opposed to "sync'ing". * Update naming convention and ensure lock for timestamp update * Add Flush method as part of WAL interface Adds a `Flush` method as part of the WAL interface to enforce the idea that we can manually trigger a WAL flush from outside of the WAL. This is employed in the consensus state management to flush the WAL prior to signing votes/proposals, as per https://github.com/tendermint/tendermint/issues/3043#issuecomment-453853630 * Update CHANGELOG_PENDING * Remove mutex approach and replace with DI The dependency injection approach to dealing with testing concerns could allow similar effects to some kind of "testing bus"-based approach. This commit introduces an example of this, where instead of relying on (potentially fragile) timing of things between the code and the test, we inject code into the function under test that can signal the test through a channel. This allows us to avoid the `time.Sleep()`-based approach previously employed. * Update comment on WAL flushing during vote signing Co-Authored-By: thanethomson <connect@thanethomson.com> * Simplify flush interval definition Co-Authored-By: thanethomson <connect@thanethomson.com> * Expand commentary on WAL disk flushing Co-Authored-By: thanethomson <connect@thanethomson.com> * Add broken test to illustrate WAL sync test problem Removes test-related state (dependency injection code) from the WAL data structure and adds test code to illustrate the problem with using `WALGenerateNBlocks` and `wal.SearchForEndHeight` to test periodic sync'ing. * Fix test error messages * Use WAL group buffer size to check for flush A function is added to `libs/autofile/group.go#Group` in order to return the size of the buffered data (i.e. data that has not yet been flushed to disk). The test now checks that, prior to a `time.Sleep`, the group buffer has data in it. After the `time.Sleep` (during which time the periodic flush should have been called), the buffer should be empty. * Remove config root dir removal from #3291 * Add godoc for NewWAL mentioning periodic sync
6 years ago
cs: sync WAL more frequently (#3300) As per #3043, this adds a ticker to sync the WAL every 2s while the WAL is running. * Flush WAL every 2s This adds a ticker that flushes the WAL every 2s while the WAL is running. This is related to #3043. * Fix spelling * Increase timeout to 2mins for slower build environments * Make WAL sync interval configurable * Add TODO to replace testChan with more comprehensive testBus * Remove extraneous debug statement * Remove testChan in favour of using system time As per https://github.com/tendermint/tendermint/pull/3300#discussion_r255886586, this removes the `testChan` WAL member and replaces the approach with a system time-oriented one. In this new approach, we keep track of the system time at which each flush and periodic flush successfully occurred. The naming of the various functions is also updated here to be more consistent with "flushing" as opposed to "sync'ing". * Update naming convention and ensure lock for timestamp update * Add Flush method as part of WAL interface Adds a `Flush` method as part of the WAL interface to enforce the idea that we can manually trigger a WAL flush from outside of the WAL. This is employed in the consensus state management to flush the WAL prior to signing votes/proposals, as per https://github.com/tendermint/tendermint/issues/3043#issuecomment-453853630 * Update CHANGELOG_PENDING * Remove mutex approach and replace with DI The dependency injection approach to dealing with testing concerns could allow similar effects to some kind of "testing bus"-based approach. This commit introduces an example of this, where instead of relying on (potentially fragile) timing of things between the code and the test, we inject code into the function under test that can signal the test through a channel. This allows us to avoid the `time.Sleep()`-based approach previously employed. * Update comment on WAL flushing during vote signing Co-Authored-By: thanethomson <connect@thanethomson.com> * Simplify flush interval definition Co-Authored-By: thanethomson <connect@thanethomson.com> * Expand commentary on WAL disk flushing Co-Authored-By: thanethomson <connect@thanethomson.com> * Add broken test to illustrate WAL sync test problem Removes test-related state (dependency injection code) from the WAL data structure and adds test code to illustrate the problem with using `WALGenerateNBlocks` and `wal.SearchForEndHeight` to test periodic sync'ing. * Fix test error messages * Use WAL group buffer size to check for flush A function is added to `libs/autofile/group.go#Group` in order to return the size of the buffered data (i.e. data that has not yet been flushed to disk). The test now checks that, prior to a `time.Sleep`, the group buffer has data in it. After the `time.Sleep` (during which time the periodic flush should have been called), the buffer should be empty. * Remove config root dir removal from #3291 * Add godoc for NewWAL mentioning periodic sync
6 years ago
cs: sync WAL more frequently (#3300) As per #3043, this adds a ticker to sync the WAL every 2s while the WAL is running. * Flush WAL every 2s This adds a ticker that flushes the WAL every 2s while the WAL is running. This is related to #3043. * Fix spelling * Increase timeout to 2mins for slower build environments * Make WAL sync interval configurable * Add TODO to replace testChan with more comprehensive testBus * Remove extraneous debug statement * Remove testChan in favour of using system time As per https://github.com/tendermint/tendermint/pull/3300#discussion_r255886586, this removes the `testChan` WAL member and replaces the approach with a system time-oriented one. In this new approach, we keep track of the system time at which each flush and periodic flush successfully occurred. The naming of the various functions is also updated here to be more consistent with "flushing" as opposed to "sync'ing". * Update naming convention and ensure lock for timestamp update * Add Flush method as part of WAL interface Adds a `Flush` method as part of the WAL interface to enforce the idea that we can manually trigger a WAL flush from outside of the WAL. This is employed in the consensus state management to flush the WAL prior to signing votes/proposals, as per https://github.com/tendermint/tendermint/issues/3043#issuecomment-453853630 * Update CHANGELOG_PENDING * Remove mutex approach and replace with DI The dependency injection approach to dealing with testing concerns could allow similar effects to some kind of "testing bus"-based approach. This commit introduces an example of this, where instead of relying on (potentially fragile) timing of things between the code and the test, we inject code into the function under test that can signal the test through a channel. This allows us to avoid the `time.Sleep()`-based approach previously employed. * Update comment on WAL flushing during vote signing Co-Authored-By: thanethomson <connect@thanethomson.com> * Simplify flush interval definition Co-Authored-By: thanethomson <connect@thanethomson.com> * Expand commentary on WAL disk flushing Co-Authored-By: thanethomson <connect@thanethomson.com> * Add broken test to illustrate WAL sync test problem Removes test-related state (dependency injection code) from the WAL data structure and adds test code to illustrate the problem with using `WALGenerateNBlocks` and `wal.SearchForEndHeight` to test periodic sync'ing. * Fix test error messages * Use WAL group buffer size to check for flush A function is added to `libs/autofile/group.go#Group` in order to return the size of the buffered data (i.e. data that has not yet been flushed to disk). The test now checks that, prior to a `time.Sleep`, the group buffer has data in it. After the `time.Sleep` (during which time the periodic flush should have been called), the buffer should be empty. * Remove config root dir removal from #3291 * Add godoc for NewWAL mentioning periodic sync
6 years ago
7 years ago
cs/wal: refuse to encode msg that is bigger than maxMsgSizeBytes (#3303) Earlier this week somebody posted this in GoS Riot chat: ``` E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 878916964 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 825701731 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 1631073634 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 912418148 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.600] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[failed to read data: EOF]" E[2019-02-12|10:38:37.600] Error on catchup replay. Proceeding to start ConsensusState anyway module=consensus err="Cannot replay height 7242. WAL does not contain #ENDHEIGHT for 7241" E[2019-02-12|10:38:37.861] Error dialing peer module=p2p err="dial tcp 35.183.126.181:26656: i/o timeout ``` Note the length error messages. What has happened is the length field got corrupted probably. I've looked at the code and noticed that we don't check the msg size during encoding. This PR fixes that. It also improves a few error messages in WALDecoder.
6 years ago
cs/wal: refuse to encode msg that is bigger than maxMsgSizeBytes (#3303) Earlier this week somebody posted this in GoS Riot chat: ``` E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 878916964 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 825701731 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 1631073634 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 912418148 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.600] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[failed to read data: EOF]" E[2019-02-12|10:38:37.600] Error on catchup replay. Proceeding to start ConsensusState anyway module=consensus err="Cannot replay height 7242. WAL does not contain #ENDHEIGHT for 7241" E[2019-02-12|10:38:37.861] Error dialing peer module=p2p err="dial tcp 35.183.126.181:26656: i/o timeout ``` Note the length error messages. What has happened is the length field got corrupted probably. I've looked at the code and noticed that we don't check the msg size during encoding. This PR fixes that. It also improves a few error messages in WALDecoder.
6 years ago
cs/wal: refuse to encode msg that is bigger than maxMsgSizeBytes (#3303) Earlier this week somebody posted this in GoS Riot chat: ``` E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 878916964 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 825701731 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 1631073634 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 912418148 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.600] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[failed to read data: EOF]" E[2019-02-12|10:38:37.600] Error on catchup replay. Proceeding to start ConsensusState anyway module=consensus err="Cannot replay height 7242. WAL does not contain #ENDHEIGHT for 7241" E[2019-02-12|10:38:37.861] Error dialing peer module=p2p err="dial tcp 35.183.126.181:26656: i/o timeout ``` Note the length error messages. What has happened is the length field got corrupted probably. I've looked at the code and noticed that we don't check the msg size during encoding. This PR fixes that. It also improves a few error messages in WALDecoder.
6 years ago
cs/wal: refuse to encode msg that is bigger than maxMsgSizeBytes (#3303) Earlier this week somebody posted this in GoS Riot chat: ``` E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 878916964 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 825701731 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 1631073634 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 912418148 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.600] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[failed to read data: EOF]" E[2019-02-12|10:38:37.600] Error on catchup replay. Proceeding to start ConsensusState anyway module=consensus err="Cannot replay height 7242. WAL does not contain #ENDHEIGHT for 7241" E[2019-02-12|10:38:37.861] Error dialing peer module=p2p err="dial tcp 35.183.126.181:26656: i/o timeout ``` Note the length error messages. What has happened is the length field got corrupted probably. I've looked at the code and noticed that we don't check the msg size during encoding. This PR fixes that. It also improves a few error messages in WALDecoder.
6 years ago
cs/wal: refuse to encode msg that is bigger than maxMsgSizeBytes (#3303) Earlier this week somebody posted this in GoS Riot chat: ``` E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 878916964 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 825701731 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 1631073634 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.596] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[length 912418148 exceeded maximum possible value of 1048576 bytes]" E[2019-02-12|10:38:37.600] Corrupted entry. Skipping... module=consensus wal=/home/gaia/.gaiad/data/cs.wal/wal err="DataCorruptionError[failed to read data: EOF]" E[2019-02-12|10:38:37.600] Error on catchup replay. Proceeding to start ConsensusState anyway module=consensus err="Cannot replay height 7242. WAL does not contain #ENDHEIGHT for 7241" E[2019-02-12|10:38:37.861] Error dialing peer module=p2p err="dial tcp 35.183.126.181:26656: i/o timeout ``` Note the length error messages. What has happened is the length field got corrupted probably. I've looked at the code and noticed that we don't check the msg size during encoding. This PR fixes that. It also improves a few error messages in WALDecoder.
6 years ago
cs: sync WAL more frequently (#3300) As per #3043, this adds a ticker to sync the WAL every 2s while the WAL is running. * Flush WAL every 2s This adds a ticker that flushes the WAL every 2s while the WAL is running. This is related to #3043. * Fix spelling * Increase timeout to 2mins for slower build environments * Make WAL sync interval configurable * Add TODO to replace testChan with more comprehensive testBus * Remove extraneous debug statement * Remove testChan in favour of using system time As per https://github.com/tendermint/tendermint/pull/3300#discussion_r255886586, this removes the `testChan` WAL member and replaces the approach with a system time-oriented one. In this new approach, we keep track of the system time at which each flush and periodic flush successfully occurred. The naming of the various functions is also updated here to be more consistent with "flushing" as opposed to "sync'ing". * Update naming convention and ensure lock for timestamp update * Add Flush method as part of WAL interface Adds a `Flush` method as part of the WAL interface to enforce the idea that we can manually trigger a WAL flush from outside of the WAL. This is employed in the consensus state management to flush the WAL prior to signing votes/proposals, as per https://github.com/tendermint/tendermint/issues/3043#issuecomment-453853630 * Update CHANGELOG_PENDING * Remove mutex approach and replace with DI The dependency injection approach to dealing with testing concerns could allow similar effects to some kind of "testing bus"-based approach. This commit introduces an example of this, where instead of relying on (potentially fragile) timing of things between the code and the test, we inject code into the function under test that can signal the test through a channel. This allows us to avoid the `time.Sleep()`-based approach previously employed. * Update comment on WAL flushing during vote signing Co-Authored-By: thanethomson <connect@thanethomson.com> * Simplify flush interval definition Co-Authored-By: thanethomson <connect@thanethomson.com> * Expand commentary on WAL disk flushing Co-Authored-By: thanethomson <connect@thanethomson.com> * Add broken test to illustrate WAL sync test problem Removes test-related state (dependency injection code) from the WAL data structure and adds test code to illustrate the problem with using `WALGenerateNBlocks` and `wal.SearchForEndHeight` to test periodic sync'ing. * Fix test error messages * Use WAL group buffer size to check for flush A function is added to `libs/autofile/group.go#Group` in order to return the size of the buffered data (i.e. data that has not yet been flushed to disk). The test now checks that, prior to a `time.Sleep`, the group buffer has data in it. After the `time.Sleep` (during which time the periodic flush should have been called), the buffer should be empty. * Remove config root dir removal from #3291 * Add godoc for NewWAL mentioning periodic sync
6 years ago
cs: sync WAL more frequently (#3300) As per #3043, this adds a ticker to sync the WAL every 2s while the WAL is running. * Flush WAL every 2s This adds a ticker that flushes the WAL every 2s while the WAL is running. This is related to #3043. * Fix spelling * Increase timeout to 2mins for slower build environments * Make WAL sync interval configurable * Add TODO to replace testChan with more comprehensive testBus * Remove extraneous debug statement * Remove testChan in favour of using system time As per https://github.com/tendermint/tendermint/pull/3300#discussion_r255886586, this removes the `testChan` WAL member and replaces the approach with a system time-oriented one. In this new approach, we keep track of the system time at which each flush and periodic flush successfully occurred. The naming of the various functions is also updated here to be more consistent with "flushing" as opposed to "sync'ing". * Update naming convention and ensure lock for timestamp update * Add Flush method as part of WAL interface Adds a `Flush` method as part of the WAL interface to enforce the idea that we can manually trigger a WAL flush from outside of the WAL. This is employed in the consensus state management to flush the WAL prior to signing votes/proposals, as per https://github.com/tendermint/tendermint/issues/3043#issuecomment-453853630 * Update CHANGELOG_PENDING * Remove mutex approach and replace with DI The dependency injection approach to dealing with testing concerns could allow similar effects to some kind of "testing bus"-based approach. This commit introduces an example of this, where instead of relying on (potentially fragile) timing of things between the code and the test, we inject code into the function under test that can signal the test through a channel. This allows us to avoid the `time.Sleep()`-based approach previously employed. * Update comment on WAL flushing during vote signing Co-Authored-By: thanethomson <connect@thanethomson.com> * Simplify flush interval definition Co-Authored-By: thanethomson <connect@thanethomson.com> * Expand commentary on WAL disk flushing Co-Authored-By: thanethomson <connect@thanethomson.com> * Add broken test to illustrate WAL sync test problem Removes test-related state (dependency injection code) from the WAL data structure and adds test code to illustrate the problem with using `WALGenerateNBlocks` and `wal.SearchForEndHeight` to test periodic sync'ing. * Fix test error messages * Use WAL group buffer size to check for flush A function is added to `libs/autofile/group.go#Group` in order to return the size of the buffered data (i.e. data that has not yet been flushed to disk). The test now checks that, prior to a `time.Sleep`, the group buffer has data in it. After the `time.Sleep` (during which time the periodic flush should have been called), the buffer should be empty. * Remove config root dir removal from #3291 * Add godoc for NewWAL mentioning periodic sync
6 years ago
cs: sync WAL more frequently (#3300) As per #3043, this adds a ticker to sync the WAL every 2s while the WAL is running. * Flush WAL every 2s This adds a ticker that flushes the WAL every 2s while the WAL is running. This is related to #3043. * Fix spelling * Increase timeout to 2mins for slower build environments * Make WAL sync interval configurable * Add TODO to replace testChan with more comprehensive testBus * Remove extraneous debug statement * Remove testChan in favour of using system time As per https://github.com/tendermint/tendermint/pull/3300#discussion_r255886586, this removes the `testChan` WAL member and replaces the approach with a system time-oriented one. In this new approach, we keep track of the system time at which each flush and periodic flush successfully occurred. The naming of the various functions is also updated here to be more consistent with "flushing" as opposed to "sync'ing". * Update naming convention and ensure lock for timestamp update * Add Flush method as part of WAL interface Adds a `Flush` method as part of the WAL interface to enforce the idea that we can manually trigger a WAL flush from outside of the WAL. This is employed in the consensus state management to flush the WAL prior to signing votes/proposals, as per https://github.com/tendermint/tendermint/issues/3043#issuecomment-453853630 * Update CHANGELOG_PENDING * Remove mutex approach and replace with DI The dependency injection approach to dealing with testing concerns could allow similar effects to some kind of "testing bus"-based approach. This commit introduces an example of this, where instead of relying on (potentially fragile) timing of things between the code and the test, we inject code into the function under test that can signal the test through a channel. This allows us to avoid the `time.Sleep()`-based approach previously employed. * Update comment on WAL flushing during vote signing Co-Authored-By: thanethomson <connect@thanethomson.com> * Simplify flush interval definition Co-Authored-By: thanethomson <connect@thanethomson.com> * Expand commentary on WAL disk flushing Co-Authored-By: thanethomson <connect@thanethomson.com> * Add broken test to illustrate WAL sync test problem Removes test-related state (dependency injection code) from the WAL data structure and adds test code to illustrate the problem with using `WALGenerateNBlocks` and `wal.SearchForEndHeight` to test periodic sync'ing. * Fix test error messages * Use WAL group buffer size to check for flush A function is added to `libs/autofile/group.go#Group` in order to return the size of the buffered data (i.e. data that has not yet been flushed to disk). The test now checks that, prior to a `time.Sleep`, the group buffer has data in it. After the `time.Sleep` (during which time the periodic flush should have been called), the buffer should be empty. * Remove config root dir removal from #3291 * Add godoc for NewWAL mentioning periodic sync
6 years ago
cs: sync WAL more frequently (#3300) As per #3043, this adds a ticker to sync the WAL every 2s while the WAL is running. * Flush WAL every 2s This adds a ticker that flushes the WAL every 2s while the WAL is running. This is related to #3043. * Fix spelling * Increase timeout to 2mins for slower build environments * Make WAL sync interval configurable * Add TODO to replace testChan with more comprehensive testBus * Remove extraneous debug statement * Remove testChan in favour of using system time As per https://github.com/tendermint/tendermint/pull/3300#discussion_r255886586, this removes the `testChan` WAL member and replaces the approach with a system time-oriented one. In this new approach, we keep track of the system time at which each flush and periodic flush successfully occurred. The naming of the various functions is also updated here to be more consistent with "flushing" as opposed to "sync'ing". * Update naming convention and ensure lock for timestamp update * Add Flush method as part of WAL interface Adds a `Flush` method as part of the WAL interface to enforce the idea that we can manually trigger a WAL flush from outside of the WAL. This is employed in the consensus state management to flush the WAL prior to signing votes/proposals, as per https://github.com/tendermint/tendermint/issues/3043#issuecomment-453853630 * Update CHANGELOG_PENDING * Remove mutex approach and replace with DI The dependency injection approach to dealing with testing concerns could allow similar effects to some kind of "testing bus"-based approach. This commit introduces an example of this, where instead of relying on (potentially fragile) timing of things between the code and the test, we inject code into the function under test that can signal the test through a channel. This allows us to avoid the `time.Sleep()`-based approach previously employed. * Update comment on WAL flushing during vote signing Co-Authored-By: thanethomson <connect@thanethomson.com> * Simplify flush interval definition Co-Authored-By: thanethomson <connect@thanethomson.com> * Expand commentary on WAL disk flushing Co-Authored-By: thanethomson <connect@thanethomson.com> * Add broken test to illustrate WAL sync test problem Removes test-related state (dependency injection code) from the WAL data structure and adds test code to illustrate the problem with using `WALGenerateNBlocks` and `wal.SearchForEndHeight` to test periodic sync'ing. * Fix test error messages * Use WAL group buffer size to check for flush A function is added to `libs/autofile/group.go#Group` in order to return the size of the buffered data (i.e. data that has not yet been flushed to disk). The test now checks that, prior to a `time.Sleep`, the group buffer has data in it. After the `time.Sleep` (during which time the periodic flush should have been called), the buffer should be empty. * Remove config root dir removal from #3291 * Add godoc for NewWAL mentioning periodic sync
6 years ago
cs: sync WAL more frequently (#3300) As per #3043, this adds a ticker to sync the WAL every 2s while the WAL is running. * Flush WAL every 2s This adds a ticker that flushes the WAL every 2s while the WAL is running. This is related to #3043. * Fix spelling * Increase timeout to 2mins for slower build environments * Make WAL sync interval configurable * Add TODO to replace testChan with more comprehensive testBus * Remove extraneous debug statement * Remove testChan in favour of using system time As per https://github.com/tendermint/tendermint/pull/3300#discussion_r255886586, this removes the `testChan` WAL member and replaces the approach with a system time-oriented one. In this new approach, we keep track of the system time at which each flush and periodic flush successfully occurred. The naming of the various functions is also updated here to be more consistent with "flushing" as opposed to "sync'ing". * Update naming convention and ensure lock for timestamp update * Add Flush method as part of WAL interface Adds a `Flush` method as part of the WAL interface to enforce the idea that we can manually trigger a WAL flush from outside of the WAL. This is employed in the consensus state management to flush the WAL prior to signing votes/proposals, as per https://github.com/tendermint/tendermint/issues/3043#issuecomment-453853630 * Update CHANGELOG_PENDING * Remove mutex approach and replace with DI The dependency injection approach to dealing with testing concerns could allow similar effects to some kind of "testing bus"-based approach. This commit introduces an example of this, where instead of relying on (potentially fragile) timing of things between the code and the test, we inject code into the function under test that can signal the test through a channel. This allows us to avoid the `time.Sleep()`-based approach previously employed. * Update comment on WAL flushing during vote signing Co-Authored-By: thanethomson <connect@thanethomson.com> * Simplify flush interval definition Co-Authored-By: thanethomson <connect@thanethomson.com> * Expand commentary on WAL disk flushing Co-Authored-By: thanethomson <connect@thanethomson.com> * Add broken test to illustrate WAL sync test problem Removes test-related state (dependency injection code) from the WAL data structure and adds test code to illustrate the problem with using `WALGenerateNBlocks` and `wal.SearchForEndHeight` to test periodic sync'ing. * Fix test error messages * Use WAL group buffer size to check for flush A function is added to `libs/autofile/group.go#Group` in order to return the size of the buffered data (i.e. data that has not yet been flushed to disk). The test now checks that, prior to a `time.Sleep`, the group buffer has data in it. After the `time.Sleep` (during which time the periodic flush should have been called), the buffer should be empty. * Remove config root dir removal from #3291 * Add godoc for NewWAL mentioning periodic sync
6 years ago
cs: sync WAL more frequently (#3300) As per #3043, this adds a ticker to sync the WAL every 2s while the WAL is running. * Flush WAL every 2s This adds a ticker that flushes the WAL every 2s while the WAL is running. This is related to #3043. * Fix spelling * Increase timeout to 2mins for slower build environments * Make WAL sync interval configurable * Add TODO to replace testChan with more comprehensive testBus * Remove extraneous debug statement * Remove testChan in favour of using system time As per https://github.com/tendermint/tendermint/pull/3300#discussion_r255886586, this removes the `testChan` WAL member and replaces the approach with a system time-oriented one. In this new approach, we keep track of the system time at which each flush and periodic flush successfully occurred. The naming of the various functions is also updated here to be more consistent with "flushing" as opposed to "sync'ing". * Update naming convention and ensure lock for timestamp update * Add Flush method as part of WAL interface Adds a `Flush` method as part of the WAL interface to enforce the idea that we can manually trigger a WAL flush from outside of the WAL. This is employed in the consensus state management to flush the WAL prior to signing votes/proposals, as per https://github.com/tendermint/tendermint/issues/3043#issuecomment-453853630 * Update CHANGELOG_PENDING * Remove mutex approach and replace with DI The dependency injection approach to dealing with testing concerns could allow similar effects to some kind of "testing bus"-based approach. This commit introduces an example of this, where instead of relying on (potentially fragile) timing of things between the code and the test, we inject code into the function under test that can signal the test through a channel. This allows us to avoid the `time.Sleep()`-based approach previously employed. * Update comment on WAL flushing during vote signing Co-Authored-By: thanethomson <connect@thanethomson.com> * Simplify flush interval definition Co-Authored-By: thanethomson <connect@thanethomson.com> * Expand commentary on WAL disk flushing Co-Authored-By: thanethomson <connect@thanethomson.com> * Add broken test to illustrate WAL sync test problem Removes test-related state (dependency injection code) from the WAL data structure and adds test code to illustrate the problem with using `WALGenerateNBlocks` and `wal.SearchForEndHeight` to test periodic sync'ing. * Fix test error messages * Use WAL group buffer size to check for flush A function is added to `libs/autofile/group.go#Group` in order to return the size of the buffered data (i.e. data that has not yet been flushed to disk). The test now checks that, prior to a `time.Sleep`, the group buffer has data in it. After the `time.Sleep` (during which time the periodic flush should have been called), the buffer should be empty. * Remove config root dir removal from #3291 * Add godoc for NewWAL mentioning periodic sync
6 years ago
  1. package consensus
  2. import (
  3. "bytes"
  4. "context"
  5. "path/filepath"
  6. "testing"
  7. "time"
  8. "github.com/stretchr/testify/assert"
  9. "github.com/stretchr/testify/require"
  10. "github.com/tendermint/tendermint/crypto/merkle"
  11. "github.com/tendermint/tendermint/internal/consensus/types"
  12. "github.com/tendermint/tendermint/internal/libs/autofile"
  13. "github.com/tendermint/tendermint/libs/log"
  14. tmtime "github.com/tendermint/tendermint/libs/time"
  15. tmtypes "github.com/tendermint/tendermint/types"
  16. )
  17. const walTestFlushInterval = 100 * time.Millisecond
  18. func TestWALTruncate(t *testing.T) {
  19. walDir := t.TempDir()
  20. walFile := filepath.Join(walDir, "wal")
  21. logger := log.TestingLogger()
  22. ctx, cancel := context.WithCancel(context.Background())
  23. defer cancel()
  24. // this magic number 4K can truncate the content when RotateFile.
  25. // defaultHeadSizeLimit(10M) is hard to simulate.
  26. // this magic number 1 * time.Millisecond make RotateFile check frequently.
  27. // defaultGroupCheckDuration(5s) is hard to simulate.
  28. wal, err := NewWAL(logger, walFile,
  29. autofile.GroupHeadSizeLimit(4096),
  30. autofile.GroupCheckDuration(1*time.Millisecond),
  31. )
  32. require.NoError(t, err)
  33. err = wal.Start(ctx)
  34. require.NoError(t, err)
  35. t.Cleanup(wal.Wait)
  36. // 60 block's size nearly 70K, greater than group's headBuf size(4096 * 10),
  37. // when headBuf is full, truncate content will Flush to the file. at this
  38. // time, RotateFile is called, truncate content exist in each file.
  39. err = WALGenerateNBlocks(ctx, t, wal.Group(), 60)
  40. require.NoError(t, err)
  41. time.Sleep(1 * time.Millisecond) // wait groupCheckDuration, make sure RotateFile run
  42. if err := wal.FlushAndSync(); err != nil {
  43. t.Error(err)
  44. }
  45. h := int64(50)
  46. gr, found, err := wal.SearchForEndHeight(h, &WALSearchOptions{})
  47. assert.NoError(t, err, "expected not to err on height %d", h)
  48. assert.True(t, found, "expected to find end height for %d", h)
  49. assert.NotNil(t, gr)
  50. t.Cleanup(func() { _ = gr.Close() })
  51. dec := NewWALDecoder(gr)
  52. msg, err := dec.Decode()
  53. assert.NoError(t, err, "expected to decode a message")
  54. rs, ok := msg.Msg.(tmtypes.EventDataRoundState)
  55. assert.True(t, ok, "expected message of type EventDataRoundState")
  56. assert.Equal(t, rs.Height, h+1, "wrong height")
  57. }
  58. func TestWALEncoderDecoder(t *testing.T) {
  59. now := tmtime.Now()
  60. msgs := []TimedWALMessage{
  61. {Time: now, Msg: EndHeightMessage{0}},
  62. {Time: now, Msg: timeoutInfo{Duration: time.Second, Height: 1, Round: 1, Step: types.RoundStepPropose}},
  63. {Time: now, Msg: tmtypes.EventDataRoundState{Height: 1, Round: 1, Step: ""}},
  64. }
  65. b := new(bytes.Buffer)
  66. for _, msg := range msgs {
  67. msg := msg
  68. b.Reset()
  69. enc := NewWALEncoder(b)
  70. err := enc.Encode(&msg)
  71. require.NoError(t, err)
  72. dec := NewWALDecoder(b)
  73. decoded, err := dec.Decode()
  74. require.NoError(t, err)
  75. assert.Equal(t, msg.Time.UTC(), decoded.Time)
  76. assert.Equal(t, msg.Msg, decoded.Msg)
  77. }
  78. }
  79. func TestWALWrite(t *testing.T) {
  80. walDir := t.TempDir()
  81. walFile := filepath.Join(walDir, "wal")
  82. ctx, cancel := context.WithCancel(context.Background())
  83. defer cancel()
  84. wal, err := NewWAL(log.TestingLogger(), walFile)
  85. require.NoError(t, err)
  86. err = wal.Start(ctx)
  87. require.NoError(t, err)
  88. t.Cleanup(wal.Wait)
  89. // 1) Write returns an error if msg is too big
  90. msg := &BlockPartMessage{
  91. Height: 1,
  92. Round: 1,
  93. Part: &tmtypes.Part{
  94. Index: 1,
  95. Bytes: make([]byte, 1),
  96. Proof: merkle.Proof{
  97. Total: 1,
  98. Index: 1,
  99. LeafHash: make([]byte, maxMsgSizeBytes-30),
  100. },
  101. },
  102. }
  103. err = wal.Write(msgInfo{
  104. Msg: msg,
  105. })
  106. if assert.Error(t, err) {
  107. assert.Contains(t, err.Error(), "msg is too big")
  108. }
  109. }
  110. func TestWALSearchForEndHeight(t *testing.T) {
  111. ctx, cancel := context.WithCancel(context.Background())
  112. defer cancel()
  113. walBody, err := WALWithNBlocks(ctx, t, 6)
  114. if err != nil {
  115. t.Fatal(err)
  116. }
  117. walFile := tempWALWithData(t, walBody)
  118. wal, err := NewWAL(log.TestingLogger(), walFile)
  119. require.NoError(t, err)
  120. h := int64(3)
  121. gr, found, err := wal.SearchForEndHeight(h, &WALSearchOptions{})
  122. assert.NoError(t, err, "expected not to err on height %d", h)
  123. assert.True(t, found, "expected to find end height for %d", h)
  124. assert.NotNil(t, gr)
  125. t.Cleanup(func() { _ = gr.Close() })
  126. dec := NewWALDecoder(gr)
  127. msg, err := dec.Decode()
  128. assert.NoError(t, err, "expected to decode a message")
  129. rs, ok := msg.Msg.(tmtypes.EventDataRoundState)
  130. assert.True(t, ok, "expected message of type EventDataRoundState")
  131. assert.Equal(t, rs.Height, h+1, "wrong height")
  132. }
  133. func TestWALPeriodicSync(t *testing.T) {
  134. walDir := t.TempDir()
  135. walFile := filepath.Join(walDir, "wal")
  136. wal, err := NewWAL(log.TestingLogger(), walFile, autofile.GroupCheckDuration(1*time.Millisecond))
  137. ctx, cancel := context.WithCancel(context.Background())
  138. defer cancel()
  139. require.NoError(t, err)
  140. wal.SetFlushInterval(walTestFlushInterval)
  141. // Generate some data
  142. err = WALGenerateNBlocks(ctx, t, wal.Group(), 5)
  143. require.NoError(t, err)
  144. // We should have data in the buffer now
  145. assert.NotZero(t, wal.Group().Buffered())
  146. require.NoError(t, wal.Start(ctx))
  147. t.Cleanup(func() {
  148. if err := wal.Stop(); err != nil {
  149. t.Error(err)
  150. }
  151. wal.Wait()
  152. })
  153. time.Sleep(walTestFlushInterval + (10 * time.Millisecond))
  154. // The data should have been flushed by the periodic sync
  155. assert.Zero(t, wal.Group().Buffered())
  156. h := int64(4)
  157. gr, found, err := wal.SearchForEndHeight(h, &WALSearchOptions{})
  158. assert.NoError(t, err, "expected not to err on height %d", h)
  159. assert.True(t, found, "expected to find end height for %d", h)
  160. assert.NotNil(t, gr)
  161. if gr != nil {
  162. gr.Close()
  163. }
  164. }