After more tests... I think that the problem is... as I optimize before every snapshot the hard links will point to completely differemt i-nodes. Once I delete the snapshots seems that Tomcat has "knid of pointers" that keep pointing to the removed snapshots so the system still thinks that shouldn't be deleted... That could explain that du command to "data" folder shows just the size of 2 index( as I just keep 2 index with an optimize after each) but df show a lot of space in use (as if no snaphost was deleted). After restaring tomcat the size after doing a df is just of 2 index, it would mean that these "tomcat pointers" have been free. Does it make sense to anyone? Does anyone know something about tomcat's way to work with hard links and disk space?? I am so confused and can't find so much info anywhere...
Thanks in advance Marc Sturlese wrote: > > Could this have something to do with the hardlinks of the snapshots? I > mean... the snapcleaner removes the snapshots but... maybe somthing > remains in there until tomcat is restarted? > > Marc Sturlese wrote: >> >> Hey there, >> Something really weird happened with my indexer... I have an index of 2G >> more or less and I am running cron jobs to keep it updated every 15 min >> (the size of the index keeps always the same more or less as I just >> update docs). Every time the update is done I optimize the index and send >> it to the slaves using the scripts of collection distribution. >> I optimize it every time because if not, searchtime becomes much more >> slow... >> The problem comes here... after 2 days of upates... tomcat broke saying >> there was no mor space left on device. Doing a df -m ... the partition >> where I have the index was at 100% but doing a du -m of the folder >> index... it was using half of the available space. >> So... I restart tomcat and everything was alrite again.. I mean ... a df >> -m was showing that just half of the partition was used. >> I have to say I just keep 2 snapshots of the index. Always delete the >> others with cronjob using snapcleaner. >> I have thought maybe it has something to do with the snapcleaner... maybe >> after delete the snapshoots tomcat still thinks that the space is in >> use...? >> I really would apreciate any advice with this because I am really lost >> with this problem... >> If I didn't explain myself well or details are missing let me know >> please... >> >> >> > > -- View this message in context: http://www.nabble.com/Really-weird-behabiour-with-the-indexer-with-cronjobs-tp22409623p22433265.html Sent from the Solr - User mailing list archive at Nabble.com.
