innobackupex does not detect lock errors
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Percona XtraBackup moved to https://jira.percona.com/projects/PXB |
Expired
|
Undecided
|
Unassigned | ||
2.0 |
Expired
|
Undecided
|
Unassigned | ||
2.1 |
Expired
|
Undecided
|
Unassigned |
Bug Description
OS: Linux db 2.6.32-
Percona-
percona-
after lock all tables error backup enters an endless loop:
>> log scanned up to (13444798386645)
120709 09:36:49 innobackupex: Continuing after ibbackup has suspended
120709 09:36:49 innobackupex: Starting mysql with options: --defaults-
120709 09:36:49 innobackupex: Connected to database with mysql child process (pid=13930)
120709 09:36:51 innobackupex: Starting to lock all tables...
>> log scanned up to (13444798539790)
innobackupex: Error: mysql child process has died: ERROR 1205 (HY000) at line 7: Lock wait timeout exceeded; try restarting transaction
while waiting for reply to MySQL request: 'FLUSH TABLES WITH READ LOCK;' at /usr//bin/
>> log scanned up to (13444798601871)
>> log scanned up to (13444798633206)
>> log scanned up to (13444798650530)
>> log scanned up to (13444798663011)
>> log scanned up to (13444798681473)
>> log scanned up to (13444798694323)
>> log scanned up to (13444798708209)
>> log scanned up to (13444798735076)
>> log scanned up to (13444798826121)
>> log scanned up to (13444798854510)
>> log scanned up to (13444798869919)
>> log scanned up to (13444798893306)
>> log scanned up to (13444798912747)
>> log scanned up to (13444799007107)
Happens to me as well:
120709 14:00:54 innobackupex: Continuing after ibbackup has suspended file='/ etc/my. cnf' --password=xxxxxxxx --user='bkpuser' --unbuffered -- innobackupex line 374.
120709 14:00:54 innobackupex: Starting mysql with options: --defaults-
120709 14:00:54 innobackupex: Connected to database with mysql child process (pid=23529)
>> log scanned up to (689 3445237648)
120709 14:00:56 innobackupex: Starting to lock all tables...
>> log scanned up to (689 3445237648)
innobackupex: Error: mysql child process has died: ERROR 1160 (08S01) at line 7: Got an error writing communication packets
while waiting for reply to MySQL request: 'FLUSH TABLES WITH READ LOCK;' at /usr/bin/
The user certainly has permissions to do the flush, and, I did so using that user within mysql client, so, I thought it might have been that but it is not.
This is with 100% innodb.