Russ Jones

Is there a specific error log where I should be able to see why it went down? It just goes down w/o any notice. It is now going down after...

Aug 17 12:40:24 dalwalrusrocksdb01 kernel: [762268.331828] ardb-server invoked oom-killer: gfp_mask=0x280da, order=0, oom_score_adj=0 Aug 17 12:40:24 dalwalrusrocksdb01 kernel: [762268.331834] ardb-server cpuset=/ mems_allowed=0 Aug 17 12:40:24 dalwalrusrocksdb01 kernel: [762268.331837] CPU: 0 PID:...

those are listings that had "ardb" listed

is there a max memory option or something that I need to set, or do I need to give the machine more RAM?

Ive got 12GB of RAM on this machine so Im really surprised it would run out.

I removed the pipeline method and am just looking everything up 1 at a time, Im watching TOP. ARDB has risen to 31.9% memory and seems to rest there nicely...

If I start requests the second I begin the ardb-server, it very rapidly progresses to 100% memory and crashes. If I wait a little bit (maybe 3 minutes), it then...

It seems like it has to be some sort of memory leak since it is occurring not because of inserts but because of hGet requests.

I will happily donate $1k to this project (if that is possible) if we can fix this problem. It is pretty mission critical at this moment.

@yinqiwen I'm not sure if this helps... [22001] 08-17 16:00:00,453 INFO coststat_hget_all:calls=300745,costs=81819213,cost_per_call=272,percents=100.0000% [22001] 08-17 16:00:00,453 INFO coststat_hget_range[0-1000]:calls=299616,costs=74902330,cost_per_call=249,percents=99.6246% [22001] 08-17 16:00:00,453 INFO coststat_hget_range[1000-5000]:calls=865,costs=1608681,cost_per_call=1859,percents=0.2876% [22001] 08-17 16:00:00,453 INFO coststat_hget_range[5000-10000]:calls=24,costs=164977,cost_per_call=6874,percents=0.0080% [22001] 08-17 16:00:00,453...