[jira] [Commented] (COUCHDB-1874) Continuous replication slows down over time
[ https://issues.apache.org/jira/browse/COUCHDB-1874?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13885276#comment-13885276 ] Benoit Chesneau commented on COUCHDB-1874: -- would be good to have a test reproducing/checking the issue. or a diff that could help. > Continuous replication slows down over time > --- > > Key: COUCHDB-1874 > URL: https://issues.apache.org/jira/browse/COUCHDB-1874 > Project: CouchDB > Issue Type: Bug > Components: Replication >Reporter: Joan Touzet > > As reported on the mailing list: > bq. > Did you specify "continuous": true in the replication spec? If you > did, do you see the same slowdown if you omit that flag? > bq. > This sounds rather similar to a bug we encountered in early builds of > BigCouch, but I don't remember the history of the codebase well enough to say > whether the bug might be latent in Apache CouchDB. > bq. Yes, I was using "continuous": true. > bq. You're right: When disabling "continuous", replication seems more stable > and efficient (1200 doc/sec). Thank you ! > And, from IRC: > bq. 12:54 <+rnewson> iirc the cause we found was an unconsumed event message > that gradually fills the mailbox, which sounds like it ought to affect > couchdb. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (COUCHDB-1874) Continuous replication slows down over time
[ https://issues.apache.org/jira/browse/COUCHDB-1874?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13885119#comment-13885119 ] ASF subversion and git services commented on COUCHDB-1874: -- Commit b35884580436824adc7f250d9adbe857440026d7 in branch refs/heads/1986-fix-ibrowse-infinite-async-timeout from [~dch] [ https://git-wip-us.apache.org/repos/asf?p=couchdb.git;h=b358845 ] ibrowse: fix async hang due to infinite timeout - import upstream timeout fix[1] - addresses COUCHDB-1986 and quite possibly COUCHDB-1874 as well [1]: https://github.com/cmullaparthi/ibrowse/pull/91.patch > Continuous replication slows down over time > --- > > Key: COUCHDB-1874 > URL: https://issues.apache.org/jira/browse/COUCHDB-1874 > Project: CouchDB > Issue Type: Bug > Components: Replication >Reporter: Joan Touzet > > As reported on the mailing list: > bq. > Did you specify "continuous": true in the replication spec? If you > did, do you see the same slowdown if you omit that flag? > bq. > This sounds rather similar to a bug we encountered in early builds of > BigCouch, but I don't remember the history of the codebase well enough to say > whether the bug might be latent in Apache CouchDB. > bq. Yes, I was using "continuous": true. > bq. You're right: When disabling "continuous", replication seems more stable > and efficient (1200 doc/sec). Thank you ! > And, from IRC: > bq. 12:54 <+rnewson> iirc the cause we found was an unconsumed event message > that gradually fills the mailbox, which sounds like it ought to affect > couchdb. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (COUCHDB-1874) Continuous replication slows down over time
[ https://issues.apache.org/jira/browse/COUCHDB-1874?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13885089#comment-13885089 ] Dave Cottlehuber commented on COUCHDB-1874: --- To be confirmed, but seems plausible. > Continuous replication slows down over time > --- > > Key: COUCHDB-1874 > URL: https://issues.apache.org/jira/browse/COUCHDB-1874 > Project: CouchDB > Issue Type: Bug > Components: Replication >Reporter: Joan Touzet > > As reported on the mailing list: > bq. > Did you specify "continuous": true in the replication spec? If you > did, do you see the same slowdown if you omit that flag? > bq. > This sounds rather similar to a bug we encountered in early builds of > BigCouch, but I don't remember the history of the codebase well enough to say > whether the bug might be latent in Apache CouchDB. > bq. Yes, I was using "continuous": true. > bq. You're right: When disabling "continuous", replication seems more stable > and efficient (1200 doc/sec). Thank you ! > And, from IRC: > bq. 12:54 <+rnewson> iirc the cause we found was an unconsumed event message > that gradually fills the mailbox, which sounds like it ought to affect > couchdb. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (COUCHDB-1874) Continuous replication slows down over time
[ https://issues.apache.org/jira/browse/COUCHDB-1874?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13819646#comment-13819646 ] Nathan Vander Wilt commented on COUCHDB-1874: - I might be seeing this with CouchDB 1.5.0 without continuous (is that likely?) replicating a decent-sized (1.5GB) database over a consumer-grade internet connection. My replication starts alright, gets another 25% through or so…then the replication more or less stops, the Erlang process starts taking 150% CPU, and it starts processing all requests very slowly. > Continuous replication slows down over time > --- > > Key: COUCHDB-1874 > URL: https://issues.apache.org/jira/browse/COUCHDB-1874 > Project: CouchDB > Issue Type: Bug > Components: Replication >Reporter: Joan Touzet > > As reported on the mailing list: > bq. > Did you specify "continuous": true in the replication spec? If you > did, do you see the same slowdown if you omit that flag? > bq. > This sounds rather similar to a bug we encountered in early builds of > BigCouch, but I don't remember the history of the codebase well enough to say > whether the bug might be latent in Apache CouchDB. > bq. Yes, I was using "continuous": true. > bq. You're right: When disabling "continuous", replication seems more stable > and efficient (1200 doc/sec). Thank you ! > And, from IRC: > bq. 12:54 <+rnewson> iirc the cause we found was an unconsumed event message > that gradually fills the mailbox, which sounds like it ought to affect > couchdb. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (COUCHDB-1874) Continuous replication slows down over time
[ https://issues.apache.org/jira/browse/COUCHDB-1874?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13745926#comment-13745926 ] Paul Hirst commented on COUCHDB-1874: - +1 I've seen exactly the same problems with continuous replication. I never tried without continuous to see if that improved things but my symptoms seem to match what's being described here. Decent speed to start with but then gradual slow down to orders of magnitude slower in a small number of hours. I ended up just restarting the database every hour until my initial replication was complete. I wish I'd just tried without continuous. > Continuous replication slows down over time > --- > > Key: COUCHDB-1874 > URL: https://issues.apache.org/jira/browse/COUCHDB-1874 > Project: CouchDB > Issue Type: Bug > Components: Replication >Reporter: Joan Touzet > > As reported on the mailing list: > bq. > Did you specify "continuous": true in the replication spec? If you > did, do you see the same slowdown if you omit that flag? > bq. > This sounds rather similar to a bug we encountered in early builds of > BigCouch, but I don't remember the history of the codebase well enough to say > whether the bug might be latent in Apache CouchDB. > bq. Yes, I was using "continuous": true. > bq. You're right: When disabling "continuous", replication seems more stable > and efficient (1200 doc/sec). Thank you ! > And, from IRC: > bq. 12:54 <+rnewson> iirc the cause we found was an unconsumed event message > that gradually fills the mailbox, which sounds like it ought to affect > couchdb. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (COUCHDB-1874) Continuous replication slows down over time
[ https://issues.apache.org/jira/browse/COUCHDB-1874?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13745155#comment-13745155 ] Joan Touzet commented on COUCHDB-1874: -- The related BigCouch change: https://github.com/cloudant/fabric/pull/40 Again from IRC: bq. <+kocolosk> the fix in Apache CouchDB master would look different I guess, but I'll bet the root cause of the process mailbox filling up with 'db_updated' messages is the same > Continuous replication slows down over time > --- > > Key: COUCHDB-1874 > URL: https://issues.apache.org/jira/browse/COUCHDB-1874 > Project: CouchDB > Issue Type: Bug > Components: Replication >Reporter: Joan Touzet > > As reported on the mailing list: > bq. > Did you specify "continuous": true in the replication spec? If you > did, do you see the same slowdown if you omit that flag? > bq. > This sounds rather similar to a bug we encountered in early builds of > BigCouch, but I don't remember the history of the codebase well enough to say > whether the bug might be latent in Apache CouchDB. > bq. Yes, I was using "continuous": true. > bq. You're right: When disabling "continuous", replication seems more stable > and efficient (1200 doc/sec). Thank you ! > And, from IRC: > bq. 12:54 <+rnewson> iirc the cause we found was an unconsumed event message > that gradually fills the mailbox, which sounds like it ought to affect > couchdb. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
