Merge lp:~akopytov/percona-xtrabackup/bug1081882-2.1 into lp:percona-xtrabackup/2.1

Proposed by Alexey Kopytov
Status: Merged
Approved by: Stewart Smith
Approved revision: no longer in the source branch.
Merged at revision: 471
Proposed branch: lp:~akopytov/percona-xtrabackup/bug1081882-2.1
Merge into: lp:percona-xtrabackup/2.1
Prerequisite: lp:~akopytov/percona-xtrabackup/bug924492-2.1
Diff against target: 13 lines (+3/-0)
1 file modified
src/fil_cur.c (+3/-0)
To merge this branch: bzr merge lp:~akopytov/percona-xtrabackup/bug1081882-2.1
Reviewer Review Type Date Requested Status
Stewart Smith (community) Approve
George Ormond Lorch III (community) g2 Approve
Review via email: mp+141919@code.launchpad.net

Description of the change

  Bug #1081882: xb_parallel_incremental test sporadically fails

  The problem was sporadic and rather rare, so one can only guess about
  the possible reasons. The most likely reason is partial page writes: if
  there is a partially written page in a block, XtraBackup retries for 10
  times to re-read before giving up.

  Naturally, it may also affects production usage, rather than just
  Jenkins builds.

  There's no ultimate fix either: regardless of how much time we allow for
  a partial page write to complete, there's always a chance it is not
  complete before we give up and claim the datafile is corrupted. However,
  spinning in a tight loop re-reading the same page(s) doesn't make much
  sense either. Given that a partial page write is supposed to be an
  extremely rare event during a backup procedure, XtraBackup can wait a
  bit longer and sleep between unsuccessful retries.

  This fix introduces a 0.1 second delay between read retries, which
  allows about 0.9 second for partial page write(s) in the same 1 MB block
  to complete.

http://jenkins.percona.com/view/XtraBackup/job/percona-xtrabackup-2.1-param/138/

To post a comment you must log in.
Revision history for this message
George Ormond Lorch III (gl-az) :
review: Approve (g2)
Revision history for this message
Stewart Smith (stewart) :
review: Approve

Preview Diff

[H/L] Next/Prev Comment, [J/K] Next/Prev File, [N/P] Next/Prev Hunk
1=== modified file 'src/fil_cur.c'
2--- src/fil_cur.c 2012-11-14 01:02:10 +0000
3+++ src/fil_cur.c 2013-01-04 14:20:36 +0000
4@@ -234,6 +234,9 @@
5 "Database page corruption detected at page "
6 "%lu, retrying...\n", cursor->thread_n,
7 page_no);
8+
9+ os_thread_sleep(100000);
10+
11 goto read_retry;
12 }
13 }

Subscribers

People subscribed via source and target branches

to all changes: