devel

devel@lists.centos.org

5 participants
4670 discussions

Back-port md raid1/10 BIO_RW_SYNC patch for DRBD
by Ross S. W. Walker 02 Mar '07

02 Mar '07

Attention CentOS Developers, Here is a back-port of Lars Ellenberg's md raid1/10 patch from 2.6.19+ to 2.6.9. It seems the md raid1 and raid10 drivers were stripping the BIO_RW_SYNC flags on requests which caused a severe performance penalty for DRBD when writing it's meta-data to these volumes. The patch is attached as well as inlined: ---------- BEGIN ---------- Subject: md: pass down BIO_RW_SYNC in raid{1,10} From: Lars Ellenberg <[EMAIL PROTECTED]> md raidX make_request functions strip off the BIO_RW_SYNC flag, thus introducing additional latency. Fixing this in raid1 and raid10 seems to be straightforward enough. For our particular usage case in DRBD, passing this flag improved some initialization time from ~5 minutes to ~5 seconds. --- raid1.c | 11 ++++++++--- raid10.c | 11 ++++++++--- 2 files changed, 16 insertions(+), 6 deletions(-) diff -puN drivers/md/raid1.c~md-pass-down-bio_rw_sync-in-raid1 drivers/md/raid1.c --- a/drivers/md/raid1.c~md-pass-down-bio_rw_sync-in-raid110 2007-02-24 01:51:20.000000000 -0500 +++ b/drivers/md/raid1.c 2007-02-24 01:50:38.000000000 -0500 @@ -521,6 +521,7 @@ static int make_request(request_queue_t struct bio *read_bio; int i, disks; mdk_rdev_t *rdev; + const int do_sync = bio_sync(bio); if (unlikely(bio_barrier(bio))) { bio_endio(bio, bio->bi_size, -EOPNOTSUPP); @@ -582,7 +583,7 @@ static int make_request(request_queue_t read_bio->bi_sector = r1_bio->sector + mirror->rdev->data_offset; read_bio->bi_bdev = mirror->rdev->bdev; read_bio->bi_end_io = raid1_end_read_request; - read_bio->bi_rw = READ; + read_bio->bi_rw = READ | do_sync; read_bio->bi_private = r1_bio; generic_make_request(read_bio); @@ -625,7 +626,7 @@ static int make_request(request_queue_t mbio->bi_sector = r1_bio->sector + conf->mirrors[i].rdev->data_offset; mbio->bi_bdev = conf->mirrors[i].rdev->bdev; mbio->bi_end_io = raid1_end_write_request; - mbio->bi_rw = WRITE; + mbio->bi_rw = WRITE | do_sync; mbio->bi_private = r1_bio; atomic_inc(&r1_bio->remaining); @@ -637,6 +638,9 @@ static int make_request(request_queue_t raid_end_bio_io(r1_bio); } + if (do_sync) + md_wakeup_thread(mddev->thread); + return 0; } @@ -960,6 +964,7 @@ static void raid1d(mddev_t *mddev) (unsigned long long)r1_bio->sector); raid_end_bio_io(r1_bio); } else { + const int do_sync = bio_sync(r1_bio->master_bio); r1_bio->bios[r1_bio->read_disk] = NULL; r1_bio->read_disk = disk; bio_put(bio); @@ -974,7 +979,7 @@ static void raid1d(mddev_t *mddev) bio->bi_sector = r1_bio->sector + rdev->data_offset; bio->bi_bdev = rdev->bdev; bio->bi_end_io = raid1_end_read_request; - bio->bi_rw = READ; + bio->bi_rw = READ | do_sync; bio->bi_private = r1_bio; unplug = 1; generic_make_request(bio); --- a/drivers/md/raid10.c~md-pass-down-bio_rw_sync-in-raid110 2007-02-24 01:51:20.000000000 -0500 +++ b/drivers/md/raid10.c 2007-02-24 01:50:38.000000000 -0500 @@ -663,6 +663,7 @@ static int make_request(request_queue_t struct bio *read_bio; int i; int chunk_sects = conf->chunk_mask + 1; + const int do_sync = bio_sync(bio); if (unlikely(bio_barrier(bio))) { bio_endio(bio, bio->bi_size, -EOPNOTSUPP); @@ -747,7 +748,7 @@ static int make_request(request_queue_t mirror->rdev->data_offset; read_bio->bi_bdev = mirror->rdev->bdev; read_bio->bi_end_io = raid10_end_read_request; - read_bio->bi_rw = READ; + read_bio->bi_rw = READ | do_sync; read_bio->bi_private = r10_bio; generic_make_request(read_bio); @@ -789,7 +790,7 @@ static int make_request(request_queue_t conf->mirrors[d].rdev->data_offset; mbio->bi_bdev = conf->mirrors[d].rdev->bdev; mbio->bi_end_io = raid10_end_write_request; - mbio->bi_rw = WRITE; + mbio->bi_rw = WRITE | do_sync; mbio->bi_private = r10_bio; atomic_inc(&r10_bio->remaining); @@ -801,6 +802,9 @@ static int make_request(request_queue_t raid_end_bio_io(r10_bio); } + if (do_sync) + md_wakeup_thread(mddev->thread); + return 0; } @@ -1247,6 +1251,7 @@ static void raid10d(mddev_t *mddev) (unsigned long long)r10_bio->sector); raid_end_bio_io(r10_bio); } else { + const int do_sync = bio_sync(r10_bio->master_bio); rdev = conf->mirrors[mirror].rdev; if (printk_ratelimit()) printk(KERN_ERR "raid10: %s: redirecting sector %llu to" @@ -1258,7 +1263,7 @@ static void raid10d(mddev_t *mddev) bio->bi_sector = r10_bio->devs[r10_bio->read_slot].addr + rdev->data_offset; bio->bi_bdev = rdev->bdev; - bio->bi_rw = READ; + bio->bi_rw = READ | do_sync; bio->bi_private = r10_bio; bio->bi_end_io = raid10_end_read_request; unplug = 1; ----------- END ----------- Ross S. W. Walker Information Systems Manager Medallion Financial, Corp. 437 Madison Avenue 38th Floor New York, NY 10022 Tel: (212) 328-2165 Fax: (212) 328-2125 WWW: http://www.medallion.com <http://www.medallion.com> ______________________________________________________________________ This e-mail, and any attachments thereto, is intended only for use by the addressee(s) named herein and may contain legally privileged and/or confidential information. If you are not the intended recipient of this e-mail, you are hereby notified that any dissemination, distribution or copying of this e-mail, and any attachments thereto, is strictly prohibited. If you have received this e-mail in error, please immediately notify the sender and permanently delete the original and any copy or printout thereof.

3 18

Nagios + selinux
by Christoph Maser 02 Mar '07

02 Mar '07

Hi lists it seems the rpmforge nagios package does not work out of the box if selinux is turned on. A log from someone complaining about it (the nagios cgis) not working: --- [Thu Mar 01 15:58:30 2007] [notice] suEXEC mechanism enabled (wrapper: /usr/sbin/suexec) [Thu Mar 01 15:58:30 2007] [notice] Digest: generating secret for digest authentication ... [Thu Mar 01 15:58:30 2007] [notice] Digest: done [Thu Mar 01 15:58:30 2007] [notice] LDAP: Built with OpenLDAP LDAP SDK [Thu Mar 01 15:58:30 2007] [notice] LDAP: SSL support unavailable [Thu Mar 01 15:58:30 2007] [notice] mod_python: Creating 4 session mutexes based on 256 max processes and 0 max threads. [Thu Mar 01 15:58:30 2007] [notice] Apache/2.0.52 (CentOS) configured -- resuming normal operations [Thu Mar 01 15:58:38 2007] [error] [client 127.0.0.1] (13)Permission denied: exec of '/usr/lib/nagios/cgi/status.cgi' failed, referer: http://127.0.0.1/nagios/side.html [Thu Mar 01 15:58:38 2007] [error] [client 127.0.0.1] Premature end of script headers: status.cgi, referer: http://127.0.0.1/nagios/side.html [Thu Mar 01 15:58:39 2007] [error] [client 127.0.0.1] (13)Permission denied: exec of '/usr/lib/nagios/cgi/tac.cgi' failed, referer: http://127.0.0.1/nagios/side.html --- I would like to make proper rules for this rpm but i have absolutely no clue about selinux and policies. Any hints what to read, where to start? Chris -- financial.com AG Tel. +49 (0) 89 / 31 85 28 - 44 Maria-Probst-Strasse 19 Fax. +49 (0) 89 / 31 85 28 - 28 D-80939 München http://www.financial.com/

6 6

CentOS 5 Beta Initial Test Release
by Hans-Joachim Hetscher 01 Mar '07

01 Mar '07

Hi, as requested by the qa-list moderator, I'd like to introduce myself: I'm working with LINUX since 1993 (I've still a almost complete floppy collection of the SLS distribution). I've done some patches for the DAMA AX25 part of the kernel. Actually, I support about 8 CentOS 4.4 and 1 Centos 3 servers. 2 of them act as a HA cluster (heartbeat + drbd) for samba. There are 2 brandnew servers waiting for CentOS5 for server consoldiation (cluster + xen). So I'd like to join the qa team, helping getting the most famous distribution. Hans-Joachim

1 0

Possible additional distribution vectors
by Daniel Clark 01 Mar '07

01 Mar '07

I was having a conversation with z00dax on IRC, and the suggestion was made that I post about this to the mailing list: CoBlitz is a nifty free distribution service for large files over http, that utilizes PlanetLab's network of hundreds of machines to provide bittorrent (or better)-like speeds while offering many of the advantages of bittorrent (less stress on mirror sites). The website is: http://codeen.cs.princeton.edu/coblitz/ To use it: ..."You add the prefix http://coblitz.codeen.org:3125/ to the URL you want to serve, and CoBlitz does the rest"... I'm sure you could do something like use tcpwrappers to limit the machines that can get to the mirror URL to the PlanetLab network. There was a good article on CoBlitz in the August 2006 edition of ;Login: - http://www.usenix.org/publications/login/2006-08/pdfs/park.pdf - however you need to be a Usenix/Sage member to read it (t becomes free August 2007). Another idea was to use a binary diff program like bsdiff - http://www.daemonology.net/bsdiff - to distribute patches to a base beta CD (that suggestion was less popular with z00dax - "thats why god made mirrors" :-) Oh, and someone might want to update the html files at http://dev.centos.org/mirrorscripts ; they reference torrents for really old beta versions. Eagerly awaiting CentOS 5 public betas, -- Daniel Clark # http://dclark.us # http://opensysadmin.com

3 2

Joining the team
by Fernando Brunelli 28 Feb '07

28 Feb '07

Hi, Just subscribed to the centos-devel list and I'd like to join devel and qa teams. I have been linux sysadmin since redhat 6, and currently running a dozen of CentOS 4 servers. I have RHCE and LPI certification, and a lot of packaging expertise. You don't know me (yet), but just give me a chance to collaborate. Thanks, Fernando Brunelli

1 0

evil centosplus dependency circle
by Michael Gabriel 28 Feb '07

28 Feb '07

trying a yum update on mysql, postgres or php forces the others to install. is that really wanted ??

7 7

centosplus php 5 dependencies
by Michael Gabriel 27 Feb '07

27 Feb '07

is it really wanted to have php 5 depend on mysql 5 AND postgres 8 ? its installing the mysql client 14 for the installed mysql 4.1.20. but why the others ? > yum update php ============================================================================= Package Arch Version Repository Size ============================================================================= Updating: php x86_64 5.1.6-3.el4s1.5 centosplus 1.2 M Installing for dependencies: mysqlclient14 x86_64 4.1.22-1.el4s1.1 centosplus 1.2M php-cli x86_64 5.1.6-3.el4s1.5 centosplus 2.3 M php-common x86_64 5.1.6-3.el4s1.5 centosplus 135 k Updating for dependencies: mysql x86_64 5.0.27-1.el4.centos centosplus 3.3 M mysql-server x86_64 5.0.27-1.el4.centos centosplus 10 M php-devel x86_64 5.1.6-3.el4s1.5 centosplus 489 k php-gd x86_64 5.1.6-3.el4s1.5 centosplus 108 k php-imap x86_64 5.1.6-3.el4s1.5 centosplus 50 k php-ldap x86_64 5.1.6-3.el4s1.5 centosplus 33 k php-mbstring x86_64 5.1.6-3.el4s1.5 centosplus 990 k php-mysql x86_64 5.1.6-3.el4s1.5 centosplus 85 k php-ncurses x86_64 5.1.6-3.el4s1.5 centosplus 39 k php-odbc x86_64 5.1.6-3.el4s1.5 centosplus 50 k php-pdo x86_64 5.1.6-3.el4s1.5 centosplus 228 k php-pgsql x86_64 5.1.6-3.el4s1.5 centosplus 66 k php-snmp x86_64 5.1.6-3.el4s1.5 centosplus 26 k php-xmlrpc x86_64 5.1.6-3.el4s1.5 centosplus 52 k postgresql x86_64 8.1.8-1.el4s1.1 centosplus 2.8 M postgresql-libs x86_64 8.1.8-1.el4s1.1 centosplus 192 k postgresql-server x86_64 8.1.8-1.el4s1.1 centosplus 4.0 M

2 1

Re: [CentOS-devel] Joining QA mailinglist
by Patrick 27 Feb '07

27 Feb '07

>> Last week, I posted my request here to join the QA team. >> Also, I signed up with the QA mailinglist. >> >> I received a confirmation after which I needed to review my settings and submit them for further processing. After that, I did not hear anything. >> >> Anyone else got the same problem? >> Should I resubmit? > >We are just being slow at approving people ... > > >Regards >Lance > Thanks for the info. I almost started to worry. Regards, Patrick _______________________________________________ No banners. No pop-ups. No kidding. Make My Way your home on the Web - http://www.myway.com

1 0

Joining QA mailinglist
by Patrick 27 Feb '07

27 Feb '07

Last week, I posted my request here to join the QA team. Also, I signed up with the QA mailinglist. I received a confirmation after which I needed to review my settings and submit them for further processing. After that, I did not hear anything. Anyone else got the same problem? Should I resubmit? Regards, Patrick _______________________________________________ No banners. No pop-ups. No kidding. Make My Way your home on the Web - http://www.myway.com

2 1

CentOS 4 Samba - Excel 2002/2003 bug
by Rodrigo Barbosa 26 Feb '07

26 Feb '07

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 There is a nasty bug on samba, up to 3.0.10, which causes Excel to incorrectly think a file was changes by someone else. This is a know issue, fixed on samba 3.0.11 (stock). A patch was also applied to samba on RHEL 3, but was NOT on RHEL 4. This patch (samba-3.0.9-excel.patch) will apply cleanly on samba-3.0.10-1.4E.9, and build without a glinch. This issue is NOT solves by http://support.microsoft.com/kb/324491 (has it on a machine with Office SP3 installed). Maybe we can get a centosplus version of samba with this patch applied ? []s - -- Rodrigo Barbosa "Quid quid Latine dictum sit, altum viditur" "Be excellent to each other ..." - Bill & Ted (Wyld Stallyns) -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (GNU/Linux) iD8DBQFFz7+rpdyWzQ5b5ckRAnX9AKCVm1x5fxyTdrjFa8kLzHLCft7tvACffTJa kpT1XyvlDjOdMoI12djA/h0= =6b5J -----END PGP SIGNATURE-----

5 6

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

devel