Hi Willi, curently the system does not swap. Because smem has a lot of dependencies I created a little script logging the statistics for arangodb (its current PID: 966) every 30 seconds: watch -n 30 "echo ==================== >> adb.log && date +%c >> adb.log && echo ==================== >> adb.log && cat /proc/966/smaps >> adb.log"
Now I would like to run a query every 30 seconds too. So I can add the results (statistics) to the log. Is there a way to pipe a query through arangosh so that I can work with the result directly on the bash? The collection size changes a little bit between the successfull attempts and the restarts. I think 50 MB or less. But this can't be the reason for the slow query. After the restart it works with the grown collections too. best regards Christoph Am Montag, 25. Juli 2016 13:26:38 UTC+2 schrieb Wilfried Gösgens: > > Hi Christoph, > can you have a look at whether the system starts swapping? > You can use smem (as found here: > http://www.cyberciti.biz/faq/linux-which-process-is-using-swap/ ) to > inspect what arangod is doing. > > The collection size doesn't change between the successfull attempts and > the restarts? > > Cheers, > Willi > > On Monday, July 25, 2016 at 10:24:59 AM UTC+2, Christoph Engel wrote: >> >> Hi, >> >> I use ArangoDB v3.0.3 under debian 8 in a LXC-container continuously >> importing some data (less than 1 MB per minute). After I've started the >> container everything runs fine. >> But after 2-3 days AQL-queries running at most 4 seconds after a restart >> of ArangoDB become very slow (Running 15 Minutes without any result). >> The data import does not seem to be affected and is still working. If I >> cancel and start a query a few times again the database crashes. >> After a restart of ArangoDB (only the DB not the LXC-container) >> everything runs fine again for the next 2-3 days. >> >> On the disc 20 GB space is available, the container has access to 6 GB >> RAM and 2 GB swap. >> The load of the machine is growing to 2 (not extremely high) and arangodb >> is the process with the highest mem-usage. >> >> I've set the logging to debug but I can't find any reason for the >> slowness of the AQL-queries or why the database crashes in the arangodb-log >> or in the systlog. >> Is there any other way to get informations about what happens to the >> arangod-deamon so that it finally crashes? >> >> best regards >> >> Christoph >> > -- You received this message because you are subscribed to the Google Groups "ArangoDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
