I have setup postgres logical replication that work well for short time. now i see that the replication status is: 'catchup', and the following error in mater log: "ERROR: could not read from reorderbuffer spill file: Not enough space" on the master drive there is 500GB free, so i assume it is not space issue any idea?
Asked
Active
Viewed 156 times
0
-
Were you closely monitoring free space **at the time the error was thrown**? – jjanes Oct 26 '19 at 16:54
-
this happen when full vacuum is running manually, any suggestion to prevent this, maybe disable logical replication prior to running vacuum? – kabanus Oct 27 '19 at 10:55
-
How big is the table you are doing the vacuum full on? If vacuum full is filling up your hard drive, and then failing and rolling back and automatically freeing up the space again, then it could cause any number of things to fail while you had a transient out of space condition. If it isn't logical replication, it will just be something else next time. (Did the vacuum full also fail?) – jjanes Oct 27 '19 at 16:10
-
we have some big size tables like 70gb and 50gb, this week we saw this error before postgres on master hangs- sqlalchemy.exc.OperationalError: (psycopg2.errors.OutOfMemory) out of memory DETAIL: Failed on request of size 48. disk space is no issue – kabanus Nov 03 '19 at 12:10
-
Those seem like independent things, out of memory and out of (disk) space. – jjanes Nov 03 '19 at 13:57
-
Thanks @jjanes. we decide to disable manual full vacuum for now. – kabanus Nov 06 '19 at 08:52