[OSM-dev] Dev is going to burn soon

Robert (Jamie) Munro rjmunro at arjam.net
Mon May 14 12:58:39 BST 2007


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Sebastian Spaeth wrote:
> Sebastian Spaeth wrote:
>>> Tile serving from dev has nearly come to a halt. the munin cpuusage stat
>>> shows iowait at 87% and load at 6.
> 
> Let me put forward an alternative theory of why that happens:
> So far we focused on hdc (the tile storage) and waited for hdd to
> provide relieve.
> 
> But looking at the IOStats on munin (remote diagnosis like that is to be
> bound to be a little crytal ball watching):
> 
> IOStat (blocks per second read/write):
> 
> Disk	Read	Write
> hdc	68	0
> hdd	33	396
> hda	499	2.2k
> ---------------------
> Total	600	2.6k
> 
> Ouch, it seems that actually hda is bearing the lions share of all reads
> (ca 83%) and writes (ca 84%). Ouch. How can that be? Apache log file
> entries being written for everything? Plus a MySQL log file entry
> written for everything? It looks like we have to give hda actually some
> relieve here. How full is hda, really?

As an experiment, could someone with access to the T at H server stuff try
changing line 4 of tiles at home's Requests/index.php so that it issues
random requests, rather than using the database. See if that reduces
load at all.

Robert (Jamie) Munro
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGSE5rz+aYVHdncI0RArlzAJ9tTvt69LOKWc1z6ymnbypXPbxBvACg1diq
FvL7wbcnWobrq+29Bk8ZyJs=
=TBle
-----END PGP SIGNATURE-----




More information about the dev mailing list