36 reputation
6
bio website
location Ukraine
age
visits member for 3 years
seen May 30 '13 at 16:46

Python programer


2d
awarded  Popular Question
Sep
24
awarded  Autobiographer
Apr
13
comment “Too many open files in system” on VPS with LSOF far from MAXFILES
Thanks for your answer! But cat /proc/sys/fs/file-nr gave me "510 0 262144" Which is even less than lsof |wc -l with current "2394". Maybe it does not work on VPS servers?
Apr
13
asked “Too many open files in system” on VPS with LSOF far from MAXFILES
Mar
27
awarded  Teacher
Mar
27
answered Web site kills hard disk I/O, how to prevent?
Feb
14
awarded  Commentator
Feb
14
comment Web site kills hard disk I/O, how to prevent?
After some tests I came to a conclusion, that this can be foult of ZEO (ZODB database). The sing is that it is designed to heavily use OS IO cache. So the problem may be that after some time of usage my App eats to much memory and the database working set does not place in disc cache. There it begins to heavily seek disk for obejcts. Will do some additional tests to find if my hunch is true.
Jan
6
comment Web site kills hard disk I/O, how to prevent?
Unfotunately - no, but the problem disapeared due to lesser load on new year. As I sayed top -m io did not bring any light on the problem. Somehow I think it can realy be related to lack of memory, because with changing number of site instances from 4 to 3 somehow fixed the issue (1GB additional memory was freed). And also we added some additional Apache image caching, so that the load would go down. Will try to reproduce the problem...
Dec
21
comment Web site kills hard disk I/O, how to prevent?
taras@XXXXXX:~> sysctl kern.coredump \n kern.coredump: 1
Dec
14
comment Web site kills hard disk I/O, how to prevent?
Speeking through comments is quite unconviniant... If you can I would prefere to use mail - taras AT sterch DOT net
Dec
14
comment Web site kills hard disk I/O, how to prevent?
* Maybe bad apache config, but It works OK for all other projects, and apache is stable as a rock, so don't know.
Dec
14
comment Web site kills hard disk I/O, how to prevent?
* I use ZODB with ZEO. ZODB: Its a database, and it's quite slow and makes many reads and writes to 1 file Data.fs. ZEO: An RPC layer around ZODB for multiple process access. Maybe many rpc client are wayting and the system becomes like this. But i know, that there is a limit to clients, so this is less posible.
Dec
14
comment Web site kills hard disk I/O, how to prevent?
Maybe there is some difference in how top -m io and gstat gather data from IO devices. Some gueses on the reason: * I have a bad cache system of images. It dumps them into 1 level of folders, so 1 folder contains around 20 000 folders with 5-6 images. Can this be a reason for this behavior.
Dec
14
comment Web site kills hard disk I/O, how to prevent?
Thanks for a good reply. Tried what you sayed, but not the result was not satisfying. I have FreeBSD 7.1-RELEASE-p5, and the top -m io realy shows me the result of IO, but it does not overlap with gstat. In other words the problem does not show itself in top -m io... Here is a simple log of script executed by cron every 2 minutes dl.dropbox.com/u/6316211/system_stat.log I truncated it to the bad period. The problem started around 17:58:00 and I restarted Apache on Wed Dec 14 18:07:22 2011. I looked at it and did not found anything particulary fishy...
Dec
14
comment Web site kills hard disk I/O, how to prevent?
Look here dl.dropbox.com/u/6316211/system_stat.log. Its a log of top -S -m io -qu -b -d 2 15. The problem occured 17.58 - 18.07
Dec
14
awarded  Supporter
Dec
14
comment Web site kills hard disk I/O, how to prevent?
Ram is 8 GB. I never saw it to go over 6. swapinfo -h Device 1K-blocks Used Avail Capacity /dev/da0s1b 4142856 24K 4.0G 0%
Dec
14
awarded  Student
Dec
9
asked Web site kills hard disk I/O, how to prevent?