glusterfs.git/geo-replication/syncdaemon/master.py, branch v3.11.3

geo-rep: Fix entry failure because parent dir doesn't exist

2017-07-10T13:59:06+00:00

In a distributed volume on master, it can so happen that
the RMDIR followed by MKDIR is recorded in changelog on
a particular subvolume with same gfid and pargfid/bname
but not on all subvolumes as below.

E 61c67a2e-07f2-45a9-95cf-d8f16a5e9c36 RMDIR \
9cc51be8-91c3-4ef4-8ae3-17596fcfed40%2Ffedora2
E 61c67a2e-07f2-45a9-95cf-d8f16a5e9c36 MKDIR 16877 0 0 \
9cc51be8-91c3-4ef4-8ae3-17596fcfed40%2Ffedora2

While processing this changelog, geo-rep thinks RMDIR is
successful and does recursive rmdir on slave. But in the
master the directory still exists. This could lead to
data discrepancy between master and slave.

Cause:
RMDIR-MKDIR pair gets recorded so in changelog when the
directory removal is successful on cached subvolume and
failed in one of hashed subvol for some reason
(may be down). In this case, the directory is re-created
on cached subvol which gets recorded as MKDIR again in
changelog.

Solution:
So while processing RMDIR geo-replication should stat on
master with gfid and should not delete it if it's present.

> Change-Id: If5da1d6462eb4d9ebe2e88b3a70cc454411a133e
> BUG: 1467718
> Signed-off-by: Kotresh HR 
> Reviewed-on: https://review.gluster.org/17695
> Smoke: Gluster Build System 
> CentOS-regression: Gluster Build System 
> Reviewed-by: Aravinda VK 
(cherry picked from commit b25bf64f3a3520a96ad557daa4903c0ceba96d72)

Change-Id: If5da1d6462eb4d9ebe2e88b3a70cc454411a133e
BUG: 1468200
Signed-off-by: Kotresh HR 
Reviewed-on: https://review.gluster.org/17715
Smoke: Gluster Build System 
CentOS-regression: Gluster Build System 
Reviewed-by: Shyamsundar Ranganathan

geo-rep: Fix xsync crawl

2017-02-16T07:50:31+00:00

If stime is set to (0, 0) on master brick root, it
is expected to do complete sync ignoring the stime
set on sub directories. But while initializing the
stime variable for comparison, it was initailized
to (-1, 0) instead of (0, 0). Fixed the same.

The stime is set to (0, 0) with the 'reset-sync-time' option
while deleting session.

'gluster vol geo-rep master fedora1::slave delete reset-sync-time'

The scenario happens when geo-rep session is deleted as above and
for some reason the session is re-established with same slave volume
after deleting data on slave volume.

Change-Id: Ie5bc8f008dead637a09495adeef5577e2b33bc90
BUG: 1422760
Signed-off-by: Kotresh HR 
Reviewed-on: https://review.gluster.org/16629
NetBSD-regression: NetBSD Build System 
CentOS-regression: Gluster Build System 
Smoke: Gluster Build System 
Reviewed-by: Aravinda VK

geo-rep: Handle directory sync failure as hard error

2017-01-13T13:00:02+00:00

If directory creation is failed, return immediately before
further processing. Allowing it to further process will
fail the entire directory tree syncing to slave. Hence
master will log and raise exception if it's directory
failure. Earlier, master used to log the failure and
proceed.

Change-Id: Iba2a8b5d3d0092e7a9c8a3c2cdf9e6e29c73ddf0
BUG: 1411607
Signed-off-by: Kotresh HR 
Reviewed-on: http://review.gluster.org/16364
Smoke: Gluster Build System 
NetBSD-regression: NetBSD Build System 
CentOS-regression: Gluster Build System 
Reviewed-by: Aravinda VK

geo-rep: Fix Last synced status column issue during Hybrid Crawl

2016-11-29T06:24:45+00:00

During Hybrid crawl, Geo-rep maintains stime xattr in subdirectories along
with the Brick root. This is done to skip directories if Geo-rep crashes
before Hybrid crawl completes.

Update Last synced status only when stime xattr updated in brick root.
Status output will mislead if it shows sub directory stime as
last synced time.

BUG: 1396081
Change-Id: I5b73aee7ae4a1c1e2d1001d1f55559b9f9efd6e6
Signed-off-by: Aravinda VK 
Reviewed-on: http://review.gluster.org/15869
NetBSD-regression: NetBSD Build System 
CentOS-regression: Gluster Build System 
Smoke: Gluster Build System 
Reviewed-by: Kotresh HR

geo-rep: Logging improvements

2016-10-24T07:01:48+00:00

- Redundant log messages removed.
- Worker and connected slave node details added in "starting worker" log
- Added log for Monitor state change
- Added log for Worker status change(Initializing/Active/Passive/Faulty)
- Added log for Crawl status Change
- Added log for config set and reset
- Added log for checkpoint set, reset and completion

BUG: 1359612
Change-Id: Icc7173ff3c93de4b862bdb1a61760db7eaf14271
Signed-off-by: Aravinda VK 
Reviewed-on: http://review.gluster.org/15684
Smoke: Gluster Build System 
NetBSD-regression: NetBSD Build System 
CentOS-regression: Gluster Build System 
Reviewed-by: Kotresh HR

geo-rep: Use configured log_level for libgfchangelog logs

2016-09-09T07:23:30+00:00

libgfchangelog was not respecting the log_level configured
in Geo-replication. With this patch Libgfchangelog log level
can be configured using `config changelog_log_level TRACE`.
Default Changelog log level is INFO

BUG: 1363965
Change-Id: Ida714931129f6a1331b9d0815da77efcb2b898e3
Signed-off-by: Aravinda VK 
Reviewed-on: http://review.gluster.org/15078
Smoke: Gluster Build System 
NetBSD-regression: NetBSD Build System 
CentOS-regression: Gluster Build System 
Reviewed-by: Kotresh HR

geo-rep: Fix History post process

2016-09-08T06:11:18+00:00

This patch removes changelogsdb part of post processing since
not got much performance advantage as expected.

Entry stime and other logging improvements retained.

BUG: 1364420
Change-Id: Ib99d23f09d96c14bc28225b47d9134260f5551bf
Signed-off-by: Aravinda VK 
Reviewed-on: http://review.gluster.org/15371
NetBSD-regression: NetBSD Build System 
CentOS-regression: Gluster Build System 
Reviewed-by: Kotresh HR 
Smoke: Gluster Build System

geo-rep: Post process Data and Meta Changelogs

2016-08-26T17:45:58+00:00

With this patch, Data and Meta GFIDs are post processed. If Changelog has
UNLINK entry then remove from Data and Meta GFIDs list(If stat on GFID is
ENOENT in Master).

While processing Changelogs,

- Collect all the data and meta operations in a temporary database
- Delete all Data and Meta GFIDs which are already unlinked as per Changelogs
  (unlink only if stat on GFID is ENOENT)
- Process all Entry operations as usual
- Process data and meta operations in batch(Fetch from Db in batch)
- Data sync is again batched based on number of changelogs(Default 1day
  changelogs). Once the sync is complete, Update last Changelog's time as last_synced
  time as usual.

Additionally maintain entry_stime on Brick root, ignore Entry ops if changelog
suffix time is less than entry_stime. If data stime is more than entry_stime,
this can happen only when passive worker updates stime by itself by getting
mount point stime. Use entry_stime = data_stime in this case.

New configurations:

max-rsync-retries - Default Value is 10
max-data-changelogs-in-batch - Max number of changelogs to be considered in a
batch for syncing. Default value is 5760(4 changelogs per min * 60 min *
24 hours)
max-history-changelogs-in-batch - Max number of history changelogs to be
processed at once. Default value 86400(4 changelogs per min * 60 min * 24
hours * 15 days)

BUG: 1364420
Change-Id: I7b665895bf4806035c2a8573d361257cbadbea17
Signed-off-by: Aravinda VK 
Reviewed-on: http://review.gluster.org/15110
Smoke: Gluster Build System 
NetBSD-regression: NetBSD Build System 
Reviewed-by: Kotresh HR 
CentOS-regression: Gluster Build System

georep: add reset-sync-time option for session delete

2016-06-29T06:41:54+00:00

Set the stime xattr at all the brick roots to (0,0) if the argument
reset-sync-time has been provided on the command-line.
To avoid testing against directory specific stime, the remote
stime is assumed to be minus_infinity, if the root directory
stime is set to (0,0), before the directory scan begins.
This triggers a full volume resync to slave in the case of a
geo-rep session recreation with the same master-slave volume
pair.

Command synopsis:
gluster volume geo-replication  :: delete \
    [reset-sync-time]

Update gluster cli man page to include new sub-command reset-sync-time.

Change-Id: Ie4ce03b9425ed9bb81eda8681058c0fc6f990948
BUG: 1311926
Signed-off-by: Milind Changire 
Reviewed-on: http://review.gluster.org/14051
Reviewed-by: Kotresh HR 
Smoke: Gluster Build System 
NetBSD-regression: NetBSD Build System 
CentOS-regression: Gluster Build System 
Reviewed-by: Aravinda VK

geo-rep: Safely handle if unliked GFID not present in data list

2016-06-20T06:37:00+00:00

If unlinked GFID is not present in data list to be synced then
Geo-rep worker was crashing with KeyError. Handled KeyError with
this patch.

BUG: 1345744
Change-Id: I5a1c9ca4473e32606df2e5c7e26c95faf55d44c0
Signed-off-by: Aravinda VK 
Reviewed-on: http://review.gluster.org/14706
Smoke: Gluster Build System 
NetBSD-regression: NetBSD Build System 
CentOS-regression: Gluster Build System 
Reviewed-by: Kotresh HR