MDB cleanup options?

300 views
Skip to first unread message

Danny Sauer

unread,
Aug 4, 2015, 6:14:10 PM8/4/15
to help-cfengine
So, we're seeing issues with large lock database files, and get this message with both 3.6.6 and 3.7.0-1 (and 3.6.{2,3,5}, though those word the error slightly differently).  I'm not certain that this is the same error manifesting, but it seems like they're probably things I shouldn't see. :)  We get this error message at the end of the run with the two newer versions:

   error: Could not advance cursor: MDB_BAD_TXN: Transaction cannot recover - it must be aborted

   Could not write database entry: MDB_BAD_TXN: Transaction cannot recover - it must be aborted


And we consistently see this after the run on the older (3.6.5 and earlier) versions, using the same system with the same big database:


   error: Could not write database entry: MDB_MAP_FULL: Environment mapsize limit reached


The lock database on these failing systems is usually 50-100MB and typically has well over half a million entries (we have a lot of promises, but not >500K).  My teammate who noticed this in testing also found issue 5637 online, which was closed.


So, this seems undesirable.

  1. should I reopen the old bug, open a new one, or try something else first?
  2. does anyone have a periodic maintenance-type command or something that we can run until the cause of this is identified?  Is there any more elegant solution than just removing the lock file after cf-agent completes successfully?  Are we really the only place seeing this? :)
Thanks.

--Danny

Nick Anderson

unread,
Aug 4, 2015, 6:52:57 PM8/4/15
to Danny Sauer, help-cfengine


On 08/04/2015 05:14 PM, Danny Sauer wrote:

And we consistently see this after the run on the older (3.6.5 and earlier)
versions, using the same system with the same big database:


error: Could not write database entry: MDB_MAP_FULL: Environment mapsize
limit reached


The lock database on these failing systems is usually 50-100MB and
typically has well over half a million entries (we have a lot of promises,
but not >500K). My teammate who noticed this in testing also found issue
5637 <https://dev.cfengine.com/issues/5637> online, which was closed.


So, this seems undesirable.


1. should I reopen the old bug, open a new one, or try something else
first?


I think you should open a new bug, and then relate it to the old one.



Reply all
Reply to author
Forward
0 new messages