Thomas Uttenthaler wrote:
here comes some dmesg output below [1]. I've discovered, that kmail (and the kmail mini icon program) also has this nasty behaviour (about every 60 seconds), of course with automatic checks for new mail deactivated. Maybe I file another bugreport about this tomorrow.
Let's take a look.
[1] korganizer alarm daemon only: yellow:/home/tom# dmesg -c korgac(4258): dirtied inode 3212122 (korgacrc.lockYIgbjb.tmp) on hda1 korgac(4258): dirtied inode 3212159 (korgacrccWoWkb.new) on hda1
An aside: Do you mount /tmp as tmpfs? That might help if these are files in /tmp.
korgac(4258): WRITE block 51413368 on hda1
Interesting. It writes it right away.
kjournald(981): WRITE block 51442208 on hda1 kjournald(981): WRITE block 51442216 on hda1 kjournald(981): WRITE block 51413360 on hda1 kjournald(981): WRITE block 16200 on hda1 kjournald(981): WRITE block 16208 on hda1 kjournald(981): WRITE block 16216 on hda1 kjournald(981): WRITE block 16224 on hda1 kjournald(981): WRITE block 16232 on hda1 kjournald(981): WRITE block 16240 on hda1 kjournald(981): WRITE block 16248 on hda1 kjournald(981): WRITE block 16256 on hda1 kjournald(981): WRITE block 16264 on hda1 kjournald(981): WRITE block 16272 on hda1 kjournald(981): WRITE block 16280 on hda1
These writes by kjournald are caused by the korgac write.
korgac(4258): dirtied inode 3213006 (?) on hda1 korgac(4258): dirtied inode 3212122 (kdeglobals.lock8f01pa.tmp) on hda1 korgac(4258): dirtied inode 3213006 (kdeglobalsrF7pzb.new) on hda1 korgac(4258): WRITE block 51441936 on hda1 korgac(4258): WRITE block 51441944 on hda1 korgac(4258): WRITE block 51442424 on hda1
It *syncs* the files, definitely! Otherwise the writes would be done by pdflush, not by korgac itself.
kjournald(981): WRITE block 51442216 on hda1 kjournald(981): WRITE block 51413384 on hda1 kjournald(981): WRITE block 16288 on hda1 kjournald(981): WRITE block 16296 on hda1 kjournald(981): WRITE block 16304 on hda1 kjournald(981): WRITE block 16312 on hda1 kjournald(981): WRITE block 16320 on hda1 kjournald(981): WRITE block 16328 on hda1 kjournald(981): WRITE block 16336 on hda1 kjournald(981): WRITE block 16344 on hda1 kjournald(981): WRITE block 16352 on hda1 kjournald(981): WRITE block 16360 on hda1 kjournald(981): WRITE block 16368 on hda1 kjournald(981): WRITE block 16376 on hda1
And a bunch of kjournald writes again, a direct result of the korgac writes. Simply the journal handling doing its thing.
korgac(4258): dirtied inode 3213022 (?) on hda1 pdflush(139): WRITE block 51442216 on hda1 pdflush(139): WRITE block 51380456 on hda1 pdflush(139): WRITE block 51413424 on hda1
These are done by pdflush however. This is a sync caused by laptop mode, I think, happening 2 seconds after disk activity ceases.
korganizer running: yellow:/home/tom# dmesg -c korgac(4258): dirtied inode 3212122 (korgacrc.lockoAlpza.tmp) on hda1 korgac(4258): dirtied inode 3212159 (korgacrcebe4fa.new) on hda1 korgac(4258): WRITE block 51441936 on hda1
korgac syncs again.
kjournald(981): WRITE block 51442256 on hda1 kjournald(981): WRITE block 51442264 on hda1 kjournald(981): WRITE block 51413360 on hda1 kjournald(981): WRITE block 19528 on hda1 kjournald(981): WRITE block 19536 on hda1 kjournald(981): WRITE block 19544 on hda1 kjournald(981): WRITE block 19552 on hda1 kjournald(981): WRITE block 19560 on hda1 kjournald(981): WRITE block 19568 on hda1 kjournald(981): WRITE block 19576 on hda1 kjournald(981): WRITE block 19584 on hda1 kjournald(981): WRITE block 19592 on hda1 kjournald(981): WRITE block 19600 on hda1 kjournald(981): WRITE block 19608 on hda1
-- indirectly caused by korgac.
korgac(4258): dirtied inode 3213006 (?) on hda1 korgac(4258): dirtied inode 3212122 (kdeglobals.lockrsh3lb.tmp) on hda1 korgac(4258): dirtied inode 3213006 (kdeglobalsNtNRxa.new) on hda1 korgac(4258): WRITE block 51441960 on hda1 korgac(4258): WRITE block 51442424 on hda1 korgac(4258): WRITE block 51442440 on hda1
korgac syncs.
kjournald(981): WRITE block 51442264 on hda1 kjournald(981): WRITE block 51441952 on hda1 kjournald(981): WRITE block 19616 on hda1 kjournald(981): WRITE block 19624 on hda1 kjournald(981): WRITE block 19632 on hda1 kjournald(981): WRITE block 19640 on hda1 kjournald(981): WRITE block 19648 on hda1 kjournald(981): WRITE block 19656 on hda1 kjournald(981): WRITE block 19664 on hda1 kjournald(981): WRITE block 19672 on hda1 kjournald(981): WRITE block 19680 on hda1 kjournald(981): WRITE block 19688 on hda1 kjournald(981): WRITE block 19696 on hda1 kjournald(981): WRITE block 19704 on hda1
caused by the previous write
korgac(4258): dirtied inode 3212191 (?) on hda1 pdflush(139): WRITE block 51442264 on hda1 pdflush(139): WRITE block 51380456 on hda1 pdflush(139): WRITE block 51413424 on hda1
This is exactly the same pattern as the last one isn't it? ;)
dmesg, kmail mini icon: yellow:/home/tom# dmesg -c kmail(4426): dirtied inode 3510175 (members) on hda1 kmail(4426): dirtied inode 3495428 (.members.index) on hda1 kmail(4426): dirtied inode 3495046 (.members.index.ids) on hda1 kmail(4426): dirtied inode 3510265 (vorstand) on hda1 kmail(4426): dirtied inode 3495045 (.vorstand.index) on hda1 kmail(4426): dirtied inode 3490104 (.vorstand.index.ids) on hda1 exim4(4602): dirtied inode 3248149 (mainlog) on hda1 exim4(4602): dirtied inode 3248149 (mainlog) on hda1 exim4(4602): dirtied inode 3248149 (mainlog) on hda1
Check that you have configured your syslog appropriately -- I mean, prepend the log file names with a '-' in syslog.conf, so that syslog doesn't sync the files immediately after every log write.
cupsd(1739): dirtied inode 3211696 (?) on hda1 cupsd(1739): dirtied inode 3211696 (0) on hda1
No writes in this output, right? So there's no spinups in this situation -- or is the log fragment too short to show it?
dmesg, kmail running: yellow:/home/tom# dmesg -c kmail(3480): dirtied inode 3344866 (Sent) on hda1 kmail(3480): dirtied inode 3345834 (.Sent.index) on hda1 kmail(3480): dirtied inode 3344869 (.Sent.index.ids) on hda1 kmail(3480): READ block 53494888 on hda1 kmail(3480): dirtied inode 3460475 (tech) on hda1 kmail(3480): dirtied inode 3346462 (.tech.index) on hda1 kmail(3480): dirtied inode 3344875 (.tech.index.ids) on hda1 kmail(3480): READ block 55328784 on hda1
It READS from disk here. That means that it's reading something that's not in cache. How long was kmail running when you got this output? It might "cool down" after it's read everything. (It might not, however.)
--Bart -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
