Multiple status files were being created by asynchronous archiving if a high-level error occurred after one or more WAL segments had already been transferred successfully. Error files were being written for every file in the queue regardless of whether it had already succeeded. To fix this, add an option to skip writing error files when an ok file already exists.
There are other situations where both files might exist (various fsync and filesystem error scenarios) so it seems best to retry in the case that multiple status files are found rather than throwing a hard error (which then means that archiving is completely stuck). In the case of multiple status files, a warning will be logged to alert the user that something unusual is happening and the command will be retried.
Reported by fpa-postgres, Joe Ayers, Douglas J Hunley.
The C info code has already been committed but this commit wires it into main.
Also remove the info Perl code and tests since they are no longer called.
Executes a child process and allows the calling process to communicate with it using read/write io.
This object is specially tailored to implement the protocol layer and may or may not be generally applicable to general purpose
execution.
Replace the repository path with just "the repository". The path is not important in this context and it is clearer to state where the stanzas are missing from.
The Perl code has a tendency to generate absolute paths even when they are not needed. This change helps the C and Perl storage work together via the protocol layer.
The C storage object strives to use rules whenever possible instead of generating absolute paths. This change helps the C and Perl storage work together via the protocol layer.
Bug Fixes:
* Remove request for S3 object info directly after putting it. (Reported by Matt Kunkel.)
* Correct archive-get-queue-max to be size type. (Reported by Ronan Dunklau.)
* Add error message when current user uid/gid does not map to a name. (Reported by Camilo Aguilar.)
* Error when --target-action=shutdown specified for PostgreSQL < 9.5.
Improvements:
* Set TCP keepalives on S3 connections. (Suggested by Ronan Dunklau.)
* Reorder info command text output so most recent backup is output last. (Contributed by Cynthia Shang. Suggested by Ryan Lambert.)
* Change file ownership only when required.
* Redact authentication header when throwing S3 errors. (Suggested by Brad Nicholson.)
Keepalives may help in situations where RST packets are being blocked by a firewall or otherwise do not arrive.
The C code uses select on all reads so it should never block, but add keepalives just in case.
Suggested by Ronan Dunklau.
After a stanza-upgrade backups for the old cluster are displayed until they expire. Cluster info was output newest to oldest which meant after an upgrade the most recent backup would no longer be output last.
Update the text output ordering so the most recent backup is always output last.
Contributed by Cynthia Shang.
Suggested by Ryan Lambert.
Previously chown() would be called even when no ownership changes were required.
In most cases changes are not required and it seems better to perform an extra stat() rather than an extra chown().
Also add unit tests for owner() since there weren't any.
The authentication header contains the access key (not the secret key) so don't include it in errors that can be seen at any log level.
Suggested by Brad Nicholson.
After a file is copied during backup the size is requested from the storage in case it differs from what was written so that repo-size can be reported accurately. This is useful for situations where compression is being done by the filesystem (e.g. ZFS) and what is stored can differ in size from what was written.
In S3 the reported size will always be exactly what was written so there is no need to check the size and doing so immediately can cause problems because the new file might not appear in list commands. This has not been observed on S3 (though it seems to be possible) but it has been reported on the Swift S3 gateway.
Add a driver capability to determine if size needs to be called after a file is written and if not then simply use the number of bytes written for repo-size.
Reported by Matt Kunkel.
Bug Fixes:
* Fix issue with archive-push-queue-max not being honored on connection error. (Reported by Lardière Sébastien.)
* Fix static WAL segment size used to determine if archive-push-queue-max has been exceeded.
* Fix error after log file open failure when processing should continue. (Reported by vthriller.)
Features:
* Automatically enable backup checksum delta when anomalies (e.g. timeline switch) are detected. (Contributed by Cynthia Shang.)
Improvements:
* Retry all S3 5xx errors rather than just 500 internal errors. (Suggested by Craig A. James.)
There are a number of cases where a checksum delta is more appropriate than the default time-based delta:
* Timeline has switched since the prior backup
* File timestamp is older than recorded in the prior backup
* File size changed but timestamp did not
* File timestamp is in the future compared to the start of the backup
* Online option has changed since the prior backup
A practical example is that checksum delta will be enabled after a failover to standby due to the timeline switch. In this case, timestamps can't be trusted and our recommendation has been to run a full backup, which can impact the retention schedule and requires manual intervention.
Now, a checksum delta will be performed if the backup type is incr/diff. This means more CPU will be used during the backup but the backup size will be smaller and the retention schedule will not be impacted.
Contributed by Cynthia Shang.
We were already retrying 500 errors but 503 (rate-limiting) errors were not being retried and would cause an instant failure which aborted the command.
There are only two 5xx errors currently implemented by S3 but instead of adding 503 simply retry all 5xx errors. This is consistent with the http definition of this error class, "the server failed to fulfill an apparently valid request."
Suggested by Craig A. James.
This calculation was missed when the WAL segment size was made dynamic in preparation for PostgreSQL 11.
Fix the calculation by checking the actual WAL file sizes instead of using an estimate based on WAL segment size. This is more accurate because it takes into account .history and .backup files, which are smaller. Since the calculation is done in the async process the additional processing time should not adversely affect performance.
Remove the PG_WAL_SIZE constant and instead use local constants where the old value is still required. This is only the case for some tests and PostgreSQL 8.3 which does not provide a way to get the WAL segment size from pg_control.
If an error occurred while acquiring a lock on a remote server the error would be reported correctly, but the queue max detection code was not reached. The tests failed to detect this because they fixed the connection before queue max, allowing the ccde to be reached.
Move the queue max code before the lock so it will run even when remote connections are not working. This means that no attempt will be made to transfer WAL once queue max has been exceeded, but it makes it much more likely that the code will be reach without error.
Update tests to continue errors up to the point where queue max is exceeded.
Reported by Lardière Sébastien.
Bug Fixes:
* Fix missing missing URI encoding in S3 driver. (Reported by Dan Farrell.)
* Fix incorrect error message for duplicate options in configuration files. (Reported by Jesper St John.)
* Fix incorrectly reported error return in info logging. A return code of 1 from the archive-get was being logged as an error message at info level but otherwise worked correctly.
Features:
* Add checksum delta for incremental backups which uses checksums rather than timestamps to determine if files have changed. (Contributed by Cynthia Shang.)
* PostgreSQL 11 support, including configurable WAL segment size.
Improvements:
* Ignore all files in a linked tablespace directory except the subdirectory for the current version of PostgreSQL. Previously an error would be generated if other files were present and not owned by the PostgreSQL user.
* Improve info command to display the stanza cipher type. (Contributed by Cynthia Shang. Suggested by Douglas J Hunley.)
* Improve support for special characters in filenames.
* Allow delta option to be specified in the pgBackRest configuration file. (Contributed by Cynthia Shang.)
PostgreSQL 11 introduces configurable WAL segment sizes, from 1MB to 1GB.
There are two areas that needed to be updated to support this: building the archive-get queue and checking that WAL has been archived after a backup. Both operations require the WAL segment size to properly build a list.
Checking the archive after a backup is still implemented in Perl and has an active database connection, so just get the WAL segment size from the database.
The archive-get command does not have a connection to the database, so get the WAL segment size from pg_control instead. This requires a deeper inspection of pg_control than has been done in the past, so it seemed best to copy the relevant data structures from each version of PostgreSQL and build a generic interface layer to address them. While this approach is a bit verbose, it has the advantage of being relatively simple, and can easily be updated for new versions of PostgreSQL.
Since the integration tests generate pg_control files for testing, teach Perl how to generate files with the correct offsets for both 32-bit and 64-bit architectures.
Use checksums rather than timestamps to determine if files have changed. This is useful in cases where the timestamps may not be trustworthy, e.g. when performing an incremental after failing over to a standby.
If checksum delta is enabled then checksums will be used for verification of resumed backups, even if they are full. Resumes have always used checksums to verify the files in the repository, enabling delta performs checksums on the database files as well.
Note that the user must manually enable this feature in cases were it would be useful or just keep in enabled all the time. A future commit will address automatically enabling the feature in cases where it seems likely to be useful.
Contributed by Cynthia Shang.
Apparently we never needed to run this function remotely.
It will be needed by the backup checksum delta feature, so implement it now.
Contributed by Cynthia Shang.
The test to make sure that some files (e.g. pg_control) do not get removed during the backup was lost during the storage refactor committed at de7fc37f.
This did not impact the integrity of the backups, but bring it back since it is a nice sanity check.
Contributed by Cynthia Shang.
The info messages were spread around and logged differently based on the execution path and in some cases logged nothing at all.
Temporarily track the async server status with a flag so that info messages are not output in the async process. The async process will be refactored as a separate command to be exec'd in a future commit.
% characters caused issues in backup/restore due to filenames being appended directly into a format string.
Reserved XML characters (<>&') caused issues in the S3 driver due to improper escaping.
Add a file with all common special characters to regression testing.
File names with uncommon characters (e.g. @) caused authentication failures due to S3 encoding them correctly while the S3 driver did not.
Reported by Dan Farrell.
A return code of 1 from the archive-get was being logged as an error message at info level but otherwise worked correctly.
Also improve info messages when an archive segment is or is not found.
Previously an error would be generated if other files were present and not owned by the PostgreSQL user. This hasn't been a big deal in practice but it could cause issues.
Also add tests to make sure the same logic applies with links to files, i.e. all other files in the directory should be ignored. This was actually working correctly, but there were no tests for it before.
Bug Fixes:
* Fix issue where relative links in $PGDATA could be stored in the backup with the wrong path. This issue did not affect absolute links and relative tablespace links were caught by other checks. (Reported by Cynthia Shang.)
* Remove incompletely implemented online option from the check command. Offline operation runs counter to the purpose of this command, which is to check if archiving and backups are working correctly. (Reported by Jason O'Donnell.)
* Fix issue where errors raised in C were not logged when called from Perl. pgBackRest properly terminated with the correct error code but lacked an error message to aid in debugging. (Reported by Douglas J Hunley.)
* Fix issue when a boolean option (e.g. delta) was specified more than once. (Reported by Yogesh Sharma.)
Features:
* Allow any option to be set in an environment variable. This includes options that previously could only be specified on the command line, e.g. stanza, and secret options that could not be specified on the command-line, e.g. repo1-s3-key-secret.
* Exclude temporary and unlogged relation (table/index) files from backup. Implemented using the same logic as the patches adding this feature to PostgreSQL, 8694cc96 and 920a5e50. Temporary relation exclusion is enabled in PostgreSQL ≥ 9.0. Unlogged relation exclusion is enabled in PostgreSQL ≥ 9.1, where the feature was introduced. (Contributed by Cynthia Shang.)
* Allow arbitrary directories and/or files to be excluded from a backup. Misuse of this feature can lead to inconsistent backups so read the --exclude documentation carefully before using. (Reviewed by Cynthia Shang.)
* Add log-subprocess option to allow file logging for local and remote subprocesses.
* PostgreSQL 11 Beta 3 support.
Improvements:
* Allow zero-size files in backup manifest to reference a prior manifest regardless of timestamp delta. (Contributed by Cynthia Shang.)
* Improve asynchronous archive-get/archive-push performance by directly checking status files. (Contributed by Stephen Frost.)
* Improve error message when a command is missing the stanza option. (Suggested by Sarah Conway.)
The log-subprocess feature added in 22765670 failed to take into account the naming for remote processes spawned by local processes. Not only was the local command used for the naming of log files but the process id was not pass through. This meant every remote log was named "[stanza]-local-remote-000" which is confusing and meant multiple processes were writing to the same log.
Instead, pass the real command and process id to the remote. This required a minor change in locking to ignore locks if process id is greater than 0 since remotes started by locals never lock.
Relative link paths were being combined with the paths of previous links (relative or absolute) due to the $strPath variable being modified in the current iteration rather than simply being passed to the next level of recursion.
This issue did not affect absolute links and relative tablespace links were caught by other checks, though the error was confusing.
Reported by Cynthia Shang.
