* Slow performance when extracting tarballs
@ 2007-04-30 12:01 Stephen So
2007-04-30 21:35 ` Chris Wedgwood
2007-04-30 23:27 ` David Chinner
0 siblings, 2 replies; 4+ messages in thread
From: Stephen So @ 2007-04-30 12:01 UTC (permalink / raw)
To: xfs
Hi everyone,
Just a question about XFS when extracting bzipped tarballs containing
lots of little files (e.g. the linux kernel source). I've noticed on my
new laptop (Intel Core 2 Duo @ 2.16 GHz), which has FC6 i386, which has
XFS partitions that when I extract these types of tarballs, my system
becomes rather non responsive, my mp3 player starts skipping etc. After
looking at top during the extraction process, bzip2 uses about 80-90% of
CPU initially and extraction seems quite fast but after a few seconds,
it drops to 30-40%, the system becomes non responsive, and extraction is
much slower.
I created a new partition of ext3 and reiserfs, did the same tarball
extraction, and on both filesystems, bzip2 uses at least 90%, extraction
is fast the whole way through, and the system is quite responsive.
I created my XFS partition using the following command (I made a larger
log file size since I heard that improve delete performance a bit):
mkfs.xfs -l size=64m /dev/sda10
Then to mount this partition, I have these switches in my /etc/fstab file
noatime, nodiratime, logbufs=8
I'm using kernel 2.6.20 that came from the FC6 updates repositories. So
is there something wrong with my XFS setup? Is my log file too small?
Or is this "normal" behaviour of XFS (i.e. that it excels best when
working with very large files but not lots of little files)?
Many thanks and best regards,
Steve.
--
______________________________________________________
Dr Stephen So, PhD, MIEEE
Griffith School of Engineering &
Institute for Integrated and Intelligent Systems
Science, Environment, Engineering and Technology Group
Griffith University, Gold Coast Campus
PMB 50, Gold Coast Mail Centre
Gold Coast, QLD, 9726, Australia
E-mail: s.so@griffith.edu.au
Phone (Fax): +61 7 5552 8663 (8065)
Homepage: http://maxwell.me.gu.edu.au/sso
______________________________________________________
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Slow performance when extracting tarballs
2007-04-30 12:01 Slow performance when extracting tarballs Stephen So
@ 2007-04-30 21:35 ` Chris Wedgwood
2007-05-04 2:45 ` Stephen So
2007-04-30 23:27 ` David Chinner
1 sibling, 1 reply; 4+ messages in thread
From: Chris Wedgwood @ 2007-04-30 21:35 UTC (permalink / raw)
To: Stephen So; +Cc: xfs
On Mon, Apr 30, 2007 at 10:01:40PM +1000, Stephen So wrote:
> After looking at top during the extraction process, bzip2 uses about
> 80-90% of CPU initially and extraction seems quite fast but after a
> few seconds, it drops to 30-40%, the system becomes non responsive,
> and extraction is much slower.
what does "vmstat 1" look like during this?
> noatime, nodiratime, logbufs=8
have you also tried setting (increasing) logbsize? (i think you need
v2 logs to make that work)
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Slow performance when extracting tarballs
2007-04-30 12:01 Slow performance when extracting tarballs Stephen So
2007-04-30 21:35 ` Chris Wedgwood
@ 2007-04-30 23:27 ` David Chinner
1 sibling, 0 replies; 4+ messages in thread
From: David Chinner @ 2007-04-30 23:27 UTC (permalink / raw)
To: Stephen So; +Cc: xfs
On Mon, Apr 30, 2007 at 10:01:40PM +1000, Stephen So wrote:
> Hi everyone,
>
> Just a question about XFS when extracting bzipped tarballs containing
> lots of little files (e.g. the linux kernel source). I've noticed on my
> new laptop (Intel Core 2 Duo @ 2.16 GHz), which has FC6 i386, which has
> XFS partitions that when I extract these types of tarballs, my system
> becomes rather non responsive, my mp3 player starts skipping etc. After
> looking at top during the extraction process, bzip2 uses about 80-90% of
> CPU initially and extraction seems quite fast but after a few seconds,
> it drops to 30-40%, the system becomes non responsive, and extraction is
> much slower.
The log probably fills up and then it falls back to the speed that
your disk can write back all the metadata.
> I created a new partition of ext3 and reiserfs, did the same tarball
> extraction, and on both filesystems, bzip2 uses at least 90%, extraction
> is fast the whole way through, and the system is quite responsive.
>
> I created my XFS partition using the following command (I made a larger
> log file size since I heard that improve delete performance a bit):
>
> mkfs.xfs -l size=64m /dev/sda10
>
> Then to mount this partition, I have these switches in my /etc/fstab file
>
> noatime, nodiratime, logbufs=8
If you don't care about the filesystem always being able to recover
correctly when power fails (i.e. can lead to filesystem coruption
on power failure) you can also use the "nobarrier" option which
can significant;y speed up metadata perfromance on XFS.
> I'm using kernel 2.6.20 that came from the FC6 updates repositories. So
> is there something wrong with my XFS setup? Is my log file too small?
> Or is this "normal" behaviour of XFS (i.e. that it excels best when
> working with very large files but not lots of little files)?
XFS excels at large files and/or lots and lots of files. On small files it
performs adequately but is not the fastest filesystem around.
Cheers,
Dave.
--
Dave Chinner
Principal Engineer
SGI Australian Software Group
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Slow performance when extracting tarballs
2007-04-30 21:35 ` Chris Wedgwood
@ 2007-05-04 2:45 ` Stephen So
0 siblings, 0 replies; 4+ messages in thread
From: Stephen So @ 2007-05-04 2:45 UTC (permalink / raw)
To: xfs
Hi, thanks for the reply
xfs-bounce@oss.sgi.com wrote:
> what does "vmstat 1" look like during this?
>
I did a vmstat 1 and this is the output:
% vmstat 1
procs -----------memory---------- ---swap-- -----io---- --system--
-----cpu------
r b swpd free buff cache si so bi bo in cs us sy
id wa st
0 0 0 1002716 3540 745316 0 0 59 15 559 560 1 2
96 1 0
0 0 0 1002700 3540 745316 0 0 0 12 1091 1543 2 2
97 0 0
1 0 0 995464 3540 750300 0 0 2060 401 1134 2569 18 3
76 4 0
2 0 0 980884 3540 762652 0 0 3712 1376 1238 4850 43 7
43 8 0
1 0 0 968368 3540 776152 0 0 3968 1568 1224 5155 43 7
44 7 0
2 0 0 954660 3540 787264 0 0 3584 1344 1244 4542 38 6
45 11 0
1 0 0 942668 3540 797556 0 0 2944 1431 1224 4376 36 6
48 11 0
1 0 0 932852 3540 807304 0 0 3072 1312 1229 4164 33 6
46 15 0
3 0 0 922724 3540 817912 0 0 3072 1440 1215 4378 37 7
44 12 0
0 1 0 911612 3540 828552 0 0 3328 1568 1242 4558 37 5
46 12 0
1 0 0 900804 3540 839140 0 0 3072 1568 1222 4279 36 5
45 13 0
0 0 0 887824 3540 848788 0 0 3072 1427 1250 3862 35 5
46 14 0
1 0 0 880036 3540 857700 0 0 2560 1529 1229 3775 31 7
47 16 0
1 0 0 867552 3540 867548 0 0 3072 1632 1250 4035 36 5
46 14 0
0 1 0 859156 3540 877576 0 0 2944 1696 1239 4291 33 6
45 16 0
1 0 0 852904 3540 883628 0 0 1664 5403 1229 3111 23 4
48 25 0
0 1 0 846328 3540 888188 0 0 1536 5300 1188 2622 21 6
61 12 0
0 1 0 842076 3540 892752 0 0 1280 5383 1232 2478 21 5
62 12 0
1 1 0 837312 3540 897396 0 0 1408 5330 1211 2476 20 5
53 24 0
6 1 0 828876 3540 903572 0 0 1920 5771 1245 2904 24 5
46 25 0
1 0 0 822016 3540 912304 0 0 2304 1203 1216 3897 30 7
55 7 0
0 1 0 818404 3540 915628 0 0 1024 9446 1181 2028 14 5
63 17 0
0 1 0 809552 3540 923336 0 0 2432 1109 1228 3344 28 5
46 22 0
1 0 0 801124 3540 928892 0 0 1664 9195 1201 2821 22 6
59 13 0
0 0 0 794364 3540 935364 0 0 1792 5296 1218 3052 24 6
52 18 0
2 1 0 789784 3540 941564 0 0 2048 4992 1194 3116 23 4
51 23 0
procs -----------memory---------- ---swap-- -----io---- --system--
-----cpu------
r b swpd free buff cache si so bi bo in cs us sy
id wa st
1 0 0 781540 3540 947180 0 0 1536 6434 1226 2942 23 6
50 22 0
4 0 0 777300 3540 953628 0 0 1920 1088 1200 2970 25 5
56 14 0
0 1 0 772892 3540 957032 0 0 1152 9440 1201 2141 17 4
59 21 0
1 0 0 764432 3540 964572 0 0 2304 1253 1216 3198 29 4
46 22 0
2 0 0 756516 3540 970284 0 0 1664 9720 1222 2832 22 5
57 17 0
1 1 0 750880 3540 977204 0 0 2176 1100 1207 2973 25 5
49 20 0
0 0 0 745424 3540 980768 0 0 1024 9140 1200 2205 16 4
66 14 0
0 1 0 741928 3540 986200 0 0 1664 1376 1193 2746 20 5
61 15 0
0 1 0 734536 3540 992480 0 0 1920 5516 1226 2874 24 5
57 14 0
0 1 0 729072 3540 997168 0 0 1408 5328 1199 2473 21 5
62 13 0
0 1 0 723228 3540 1003288 0 0 1792 5509 1243 2959 24 6
54 15 0
2 0 0 717948 3540 1007752 0 0 1408 5308 1196 2418 20 4
59 18 0
4 0 0 709940 3540 1013564 0 0 1536 5568 1217 3145 25 4
55 16 0
0 0 0 701132 3540 1021948 0 0 2816 5612 1224 3562 32 6
47 16 0
0 1 0 702448 3540 1023140 0 0 256 5108 1203 1538 6 5
73 15 0
0 1 0 691688 3540 1032264 0 0 2688 1852 1239 3630 32 5
45 18 0
0 1 0 688292 3540 1034228 0 0 768 9348 1198 1671 10 3
60 27 0
1 0 0 682636 3540 1039248 0 0 1408 1069 1198 2729 20 5
47 29 0
1 0 0 676848 3540 1044456 0 0 1408 5704 1234 2897 20 5
59 16 0
1 0 0 672460 3540 1049428 0 0 1536 5484 1215 2813 19 5
55 22 0
1 0 0 663820 3540 1056108 0 0 2176 5258 1241 3245 27 5
49 20 0
1 0 0 660064 3540 1061708 0 0 1664 1688 1222 3100 22 6
60 11 0
0 0 0 653400 3540 1065924 0 0 1152 5496 1221 2495 17 4
51 28 0
0 1 0 651468 3540 1069324 0 0 1152 5278 1187 2157 16 3
67 14 0
2 0 0 645132 3540 1073620 0 0 1152 5466 1221 2714 19 5
61 17 0
3 0 0 640544 3540 1078720 0 0 1664 5587 1219 2830 21 6
51 21 0
procs -----------memory---------- ---swap-- -----io---- --system--
-----cpu------
r b swpd free buff cache si so bi bo in cs us sy
id wa st
1 0 0 634040 3540 1083872 0 0 1536 5223 1208 2996 20 3
64 14 0
0 1 0 629024 3540 1090772 0 0 2048 5342 1199 3141 26 5
49 20 0
0 0 0 621116 3540 1095840 0 0 1664 4410 1211 2631 22 4
52 22 0
0 0 0 615760 3540 1100840 0 0 1408 6032 1186 2601 20 6
61 14 0
0 0 0 608852 3540 1107448 0 0 1920 1192 1215 3228 24 6
50 21 0
0 1 0 605872 3540 1112248 0 0 1536 5424 1220 2779 22 4
63 12 0
0 0 0 598016 3540 1117476 0 0 1536 5603 1227 3016 23 4
53 21 0
2 0 0 592416 3540 1122576 0 0 1536 5407 1217 2671 22 7
56 16 0
0 0 0 587504 3540 1127404 0 0 1408 4624 1230 2599 19 5
55 21 0
2 1 0 585800 3540 1130704 0 0 1152 1880 1175 2431 15 2
53 30 0
0 1 0 582732 3540 1133696 0 0 896 5293 1210 2357 16 4
74 6 0
2 0 0 575528 3540 1138696 0 0 1536 5424 1214 2585 22 5
48 26 0
1 0 0 569992 3540 1145872 0 0 2176 1519 1245 3267 27 5
50 17 0
0 0 0 563568 3540 1149772 0 0 1152 8164 1189 2364 15 4
74 6 0
0 0 0 559936 3540 1153020 0 0 896 2483 1198 2145 16 6
64 15 0
1 0 0 556504 3540 1156720 0 0 1408 5248 1206 2152 17 6
62 14 0
0 1 0 553568 3540 1161280 0 0 1280 5716 1231 2620 19 4
59 18 0
1 0 0 544820 3540 1167580 0 0 2048 1545 1234 2947 26 5
51 18 0
1 0 0 541096 3540 1170748 0 0 1024 5272 1205 2107 17 3
73 8 0
0 1 0 535092 3540 1176848 0 0 1792 6132 1225 2861 25 6
49 19 0
0 1 0 531696 3540 1181220 0 0 1280 969 1215 2758 18 3
66 14 0
0 1 0 528920 3540 1184220 0 0 896 5268 1192 2248 16 4
71 10 0
0 0 0 520532 3540 1189884 0 0 1664 5425 1252 3008 21 4
64 10 0
0 0 0 514012 3540 1196084 0 0 1920 1920 1214 3110 25 6
60 10 0
1 0 0 511804 3540 1199608 0 0 1152 5336 1240 2224 20 6
60 15 0
0 0 0 503212 3540 1206108 0 0 2048 6516 1227 2963 26 4
58 12 0
procs -----------memory---------- ---swap-- -----io---- --system--
-----cpu------
r b swpd free buff cache si so bi bo in cs us sy
id wa st
2 0 0 500968 3540 1208380 0 0 512 4684 1214 2066 13 5
79 3 0
2 0 0 496216 3540 1212580 0 0 1408 5727 1214 2399 21 7
65 8 0
4 0 0 491268 3540 1217184 0 0 1408 4304 1243 2593 21 5
64 11 0
2 0 0 488856 3540 1219784 0 0 896 2058 1189 1849 15 4
70 11 0
0 1 0 483660 3540 1224340 0 0 1408 5824 1240 2571 23 4
52 22 0
0 0 0 477704 3540 1229940 0 0 1536 5170 1173 2855 21 6
52 21 0
0 1 0 474500 3540 1234952 0 0 1536 5163 1212 2629 20 3
55 23 0
1 0 0 465196 3540 1242552 0 0 2304 940 1204 3265 28 4
47 22 0
0 0 0 458280 3540 1247892 0 0 1664 9382 1211 2719 19 6
70 4 0
1 0 0 453276 3540 1252592 0 0 1408 5040 1176 2827 19 5
58 18 0
0 0 0 446840 3540 1258496 0 0 1792 5676 1221 3025 24 5
56 14 0
1 0 0 443180 3540 1264096 0 0 1664 932 1193 2680 21 5
56 19 0
1 0 0 435748 3540 1269060 0 0 1664 5182 1209 2635 21 4
49 27 0
0 1 0 432060 3540 1274860 0 0 1536 5376 1183 2860 21 7
51 20 0
0 1 0 426376 3540 1279492 0 0 1408 5177 1214 2480 19 3
51 27 0
0 1 0 422356 3540 1283992 0 0 1280 5256 1196 2516 18 4
55 23 0
0 0 0 410112 3540 1292848 0 0 2560 1916 1254 3839 32 5
49 14 0
1 0 0 407296 3540 1295448 0 0 896 8244 1203 1816 15 6
74 6 0
1 0 0 405256 3540 1297456 0 0 384 2276 1192 1729 10 5
78 7 0
1 0 0 401044 3540 1303756 0 0 1920 2004 1260 2779 29 5
56 11 0
1 0 0 397668 3540 1306976 0 0 1024 5432 1229 2264 18 4
68 9 0
1 0 0 393720 3540 1310076 0 0 1024 5520 1219 1983 17 6
66 11 0
1 0 0 384148 3540 1316896 0 0 2048 2224 1279 3279 33 5
54 9 0
1 0 0 384716 3540 1318996 0 0 336 5291 1194 2084 12 3
74 13 0
0 0 0 384716 3540 1319252 0 0 0 149 1115 1467 1 1
98 0 0
0 0 0 384716 3540 1319252 0 0 0 92 1065 1075 2 3
95 0 0
> have you also tried setting (increasing) logbsize? (i think you need
>
> v2 logs to make that work)
>
I read in the man page for mount that the max logbsize is 32K and the
default value for machines with more than 32 MB of memory is 32768, so I
assumed it was already set to maximum.
Best regards,
Steve.
--
__________________________________________________
Dr Stephen So, PhD, MIEEE
Griffith School of Engineering &
Institute for Integrated and Intelligent Systems
Griffith University, Gold Coast Campus
PMB 50 Gold Coast Mail Centre
Gold Coast, QLD, 9726, Australia.
E-mail: s.so@griffith.edu.au
Phone: +61 7 5552 8663
Fax: +61 7 5552 8065
__________________________________________________
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2007-05-04 2:46 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2007-04-30 12:01 Slow performance when extracting tarballs Stephen So
2007-04-30 21:35 ` Chris Wedgwood
2007-05-04 2:45 ` Stephen So
2007-04-30 23:27 ` David Chinner
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox