> Scott, can you build a test package with this patch for Tore?
>
> So Tore can test, if it makes the timing acceptable on his box.
Don't bother with that... It should be enough to just test vol_id:
root@cuba:~# time /tmp/vol_id-udev113orig --export /dev/sdk
/dev/sdk: unknown volume type
real 2m1.113s
user 0m0.000s
sys 0m0.000s
root@cuba:~# time /tmp/vol_id-udev113patched --export /dev/sdk
/dev/sdk: unknown volume type
real 0m6.047s
user 0m0.000s
sys 0m0.000s
It's MUCH better with the patch. However I think it is possible to make
it even twice as quick, since it generates two I/O errors still:
Dec 17 09:36:13 cuba kernel: [3942923.936191] end_request: I/O error, dev sdk, sector 0
Dec 17 09:36:13 cuba kernel: [3942923.946615] Buffer I/O error on device sdk, logical block 0
Dec 17 09:36:16 cuba kernel: [3942926.955903] end_request: I/O error, dev sdk, sector 0
Dec 17 09:36:16 cuba kernel: [3942926.966328] Buffer I/O error on device sdk, logical block 0
In comparsion "fdisk -l /dev/sdk" takes 3 seconds and only one such
error is logged per invocation.
I can also confirm that the patch does not appear to break anything for
reachable devices:
* Kay Sievers
> Scott, can you build a test package with this patch for Tore?
>
> So Tore can test, if it makes the timing acceptable on his box.
Don't bother with that... It should be enough to just test vol_id:
root@cuba:~# time /tmp/vol_ id-udev113orig --export /dev/sdk
/dev/sdk: unknown volume type
real 2m1.113s id-udev113patch ed --export /dev/sdk
user 0m0.000s
sys 0m0.000s
root@cuba:~# time /tmp/vol_
/dev/sdk: unknown volume type
real 0m6.047s
user 0m0.000s
sys 0m0.000s
It's MUCH better with the patch. However I think it is possible to make
it even twice as quick, since it generates two I/O errors still:
Dec 17 09:36:13 cuba kernel: [3942923.936191] end_request: I/O error, dev sdk, sector 0
Dec 17 09:36:13 cuba kernel: [3942923.946615] Buffer I/O error on device sdk, logical block 0
Dec 17 09:36:16 cuba kernel: [3942926.955903] end_request: I/O error, dev sdk, sector 0
Dec 17 09:36:16 cuba kernel: [3942926.966328] Buffer I/O error on device sdk, logical block 0
In comparsion "fdisk -l /dev/sdk" takes 3 seconds and only one such
error is logged per invocation.
I can also confirm that the patch does not appear to break anything for
reachable devices:
root@cuba:~# /tmp/vol_ id-udev113orig --export /dev/sde filesystem df6f538e- a8b4-4195- 91e9-4989cb6bfb 45 ENC=df6f538e- a8b4-4195- 91e9-4989cb6bfb 45 honolulu/ mysql ENC=honolulu\ x2fmysql SAFE=honolulumy sql id-udev113patch ed --export /dev/sde filesystem df6f538e- a8b4-4195- 91e9-4989cb6bfb 45 ENC=df6f538e- a8b4-4195- 91e9-4989cb6bfb 45 honolulu/ mysql ENC=honolulu\ x2fmysql SAFE=honolulumy sql
ID_FS_USAGE=
ID_FS_TYPE=ext3
ID_FS_VERSION=1.0
ID_FS_UUID=
ID_FS_UUID_
ID_FS_LABEL=
ID_FS_LABEL_
ID_FS_LABEL_
root@cuba:~# /tmp/vol_
ID_FS_USAGE=
ID_FS_TYPE=ext3
ID_FS_VERSION=1.0
ID_FS_UUID=
ID_FS_UUID_
ID_FS_LABEL=
ID_FS_LABEL_
ID_FS_LABEL_
Regards
--
Tore Anderson