Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Conversation

timvaillancourt
Copy link
Contributor

@timvaillancourt timvaillancourt commented Jul 21, 2017

  1. GCS (Google Cloud Storage) upload method w/threading (1.0.4: Support Google Cloud Storage upload v1 #178, Readme gcs #181, Readme gcs 2 #182, 1.1.0 gcs upload threading #185).
  2. Fabric and Boto requirements updates (to latest) (1.1.0 requirements update #183).
  3. New feature to disable cluster oplog tailing (ie: cluster-wide consistency) (1.0.4: Fix disabling of oplog tailer #175).
  4. Fsync mechanism added to oplog tailing, 1 x fsync every 1 sec or 100 oplog writes (by default) (1.0.4: Support handling of TailThread cursor/connection failures #174).
  5. Allow failures in oplog tailing to stop backup and dump method (1.0.4: Support handling of TailThread cursor/connection failures #174).
  6. Ensure Resolver consistent timestamp is not before end of mongodump (Issue The backup is not point in time if a tailed oplog's last timestamp is before a mongodump's. #164, fix: 1.0.4: Ensure Resolver consistent timestamp is not before end of mongodump (#164) #171).
  7. Only apply quorum checking to electable nodes (1.0.4: Remove 'secondary_count' logic, only count electable nodes in quorum #169).
  8. Default replication delay maximum raised to 10 secs from 5.
  9. Display more mongodump version information (1.0.4: Print more mongodump details #170).
  10. Handle mongodump errors properly (1.0.4: Mongodump handle error #176).
  11. Hide GCS secret keys in logging.
  12. 'make docker' must depend on the bin/mongodb-consistent-backup build in Makefile (1.0.4: Mongodump handle error #176).

1.0.4: Only consider electable members for replset quorum count
* Print more details about mongodump

* Version must be set first

* Check if version unknown
…odump (#164) (#171)

* Don't return an consistent_end_ts that is earlier than the last_ts of mongodump oplogs
* Some code cleanup and fixed rounding-up logic
Support better handling/logging of TailThread cursor/connection failures
Fsync the tailed oplog to disk on time or doc-writes thresholds.
Pass 'backup_stop' event to threads to signal failure to other child threads.
Allow oplog tailing to be disabled.
Added script to simulate failure of cursor/query.
* Handle mongodump's "Failed: ..." messages as a proper error
* Update Fabric and boto versions
* Raise default repl lag max to 10s, 5s has shown to be too low on many hosts. 3.4 adds a no-op oplog update every 10s so this seems to align nicely with it
@timvaillancourt timvaillancourt requested a review from dbmurphy July 21, 2017 14:14
@timvaillancourt timvaillancourt added this to the 1.1.0 milestone Jul 21, 2017
@timvaillancourt timvaillancourt self-assigned this Jul 21, 2017
@dbmurphy
Copy link
Contributor

dbmurphy commented Aug 4, 2017

In the future, we should make PR's to master for each thing we do, then just cut a version for release, these mega merges are hard to trust :)

@timvaillancourt timvaillancourt merged commit 34818a2 into master Aug 4, 2017
@timvaillancourt timvaillancourt deleted the 1.1.0 branch September 1, 2017 15:59
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants