Skip to contentSkip to conversation

dmeventd_mirror problem and some questions.

linux-lvm Simone Gotti 2 participants 4 messages Apr 18, 2006 – Apr 21, 2006

lore
Share & tools
lore
#0 lore raw

Hi All,

I was curious to try the experimental dmeventd + the mirror checking dso.
Everytime I created a mirror volume, dmeventd was crashing when the mirror finished the syncing process.
I tracked it down to _get_mirror_event that is expeting a return string from the status ioctl to dm like "2 253:4 253:5 400/400 1 AA 3 cluster 253:3 A" but it gets only "2 253:4 253:5 400/400" so it SIGSEGV when is going to read over the number of returned arguments in the "args" array from split_words.

The same string is returned by "dmsetup status".

Do these informations aren't already implemented in the ioctl, or are there some patches around that I can try? (my kernel is the latest available in fedora rawhide: 2.6.16-1.2131_FC6)

Another question, If I understand it correctly, the mirror library for dmeventd will look to possible problems (eg the lost of a PV containing a mirror), and it will call the right lvm command to fix it? For example, if a PV containing a side of a 2 way mirrored volume is lost, what are the various actions that it will do (I noticed an lvreduce --removemissing, but my fear is that this command will also remove unmirrored volumes)?

Thanks!

Bye!

Simone Gotti.

--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
Problemi di Liquidit�? Con Logos Finanziaria 30.000 � in 24 ore a dipendenti e lavoratori autonomi con rimborsi fino a 120 mesi, clicca qui
*
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=2911&d=18-4

#1 lore raw

There are some kernel patches that you need. Either way, the DSO shouldn't be crashing, so I'll look into that.

Kernel patches are located at http://www.brassow.com/mirroring. Last updated on April 3rd... I'll be updating those either today or tomorrow as well.

brassow

Quoted 51 lines · Simone Gotti

Hi All,

I was curious to try the experimental dmeventd + the mirror checking dso.
Everytime I created a mirror volume, dmeventd was crashing when the mirror finished the syncing process.
I tracked it down to _get_mirror_event that is expeting a return string from the status ioctl to dm like "2 253:4 253:5 400/400 1 AA 3 cluster 253:3 A" but it gets only "2 253:4 253:5 400/400" so it SIGSEGV when is going to read over the number of returned arguments in the "args" array from split_words.

The same string is returned by "dmsetup status".

Do these informations aren't already implemented in the ioctl, or are there some patches around that I can try? (my kernel is the latest available in fedora rawhide: 2.6.16-1.2131_FC6)

Another question, If I understand it correctly, the mirror library for dmeventd will look to possible problems (eg the lost of a PV containing a mirror), and it will call the right lvm command to fix it? For example, if a PV containing a side of a 2 way mirrored volume is lost, what are the various actions that it will do (I noticed an lvreduce --removemissing, but my fear is that this command will also remove unmirrored volumes)?

Thanks!

Bye!

Simone Gotti.

--
Email.it, the professional e-mail, gratis per te:
http://www.email.it/f

Sponsor:
Problemi di Liquidit�? Con Logos Finanziaria 30.000 � in 24 ore a
dipendenti e lavoratori autonomi con rimborsi fino a 120 mesi, clicca qui *
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=2911&d=18-4

_______________________________________________
linux-lvm mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

#2 lore raw
Quoted 3 lines · Jonathan E Brassow

Kernel patches are located at http://www.brassow.com/mirroring. Last updated on April 3rd... I'll be updating those either today or tomorrow as well.

Thanks,
I tried the version of the 3 April and the new version of yesterday. I applied the patches in the order provided by the "series" file, used the standard fedora .config, but on both the versions I got this compilation error:

CC [M] drivers/md/dm-log.o
drivers/md/dm-log.c: In function ‘dm_register_dirty_log_type’:
drivers/md/dm-log.c:19: error: ‘_lock’ undeclared (first use in this function)
drivers/md/dm-log.c:19: error: (Each undeclared identifier is reported only once
drivers/md/dm-log.c:19: error: for each function it appears in.) drivers/md/dm-log.c: In function ‘dm_unregister_dirty_log_type’:
drivers/md/dm-log.c:29: error: ‘_lock’ undeclared (first use in this function)
drivers/md/dm-log.c: In function ‘get_type’:
drivers/md/dm-log.c:46: error: ‘_lock’ undeclared (first use in this function)
drivers/md/dm-log.c: In function ‘put_type’:
drivers/md/dm-log.c:64: error: ‘_lock’ undeclared (first use in this function)
make[2]: *** [drivers/md/dm-log.o] Error 1
make[1]: *** [drivers/md] Error 2
make: *** [drivers] Error 2

I am missing something?

Quoted 2 lines

There are some kernel patches that you need. Either way, the DSO shouldn't be crashing, so I'll look into that.

My temporary workaround was to check the number of returned arguments:

Index: dmeventd_mirror.c
===================================================================
RCS file: /cvs/lvm2/LVM2/dmeventd/mirror/dmeventd_mirror.c,v retrieving revision 1.5

dmeventd_mirror.c 20 Apr 2006 20:42:18 -0000 renamed from dmeventd_mirror.c 6 Feb 2006 19:34:45 -0000 1.5 +3 -1
diff -u -b -B -p -r1.5 dmeventd_mirror.c
@@ -39,6 +39,7 @@ static int _get_mirror_event(char *param
 {
        int i, rtn = ME_INSYNC;
        int max_args = 30;  /* should support at least 8-way mirrors */
+       int output_args;
        char *args[max_args];
        char *dev_status_str;
        char *log_status_str;
@@ -46,11 +47,17 @@ static int _get_mirror_event(char *param
        char *p;
        int log_argc, num_devs, num_failures=0;
 
-       if (max_args <= split_words(params, max_args, args)) {
+       output_args = split_words(params, max_args, args);
+       if (max_args <= output_args) {
                syslog(LOG_ERR, "Unable to split mirror parameters: Arg
list too long");
                return -E2BIG;
        }

+ if (output_args <= 4) {
+ syslog(LOG_ERR, "Mirror status not provided by the
kernel module");
+ return ME_IGNORE;
+ };
+
/*
* Unused: 0 409600 mirror
* Used : 2 253:4 253:5 400/400 1 AA 3 cluster 253:3 A

Thanks!

Bye!

Quoted 58 lines

brassow

Quoted 46 lines · Simone Gotti

Hi All,

I was curious to try the experimental dmeventd + the mirror checking dso.
Everytime I created a mirror volume, dmeventd was crashing when the mirror finished the syncing process.
I tracked it down to _get_mirror_event that is expeting a return string from the status ioctl to dm like "2 253:4 253:5 400/400 1 AA 3 cluster 253:3 A" but it gets only "2 253:4 253:5 400/400" so it SIGSEGV when is going to read over the number of returned arguments in the "args" array from split_words.

The same string is returned by "dmsetup status".

Do these informations aren't already implemented in the ioctl, or are there some patches around that I can try? (my kernel is the latest available in fedora rawhide: 2.6.16-1.2131_FC6)

Another question, If I understand it correctly, the mirror library for dmeventd will look to possible problems (eg the lost of a PV containing a mirror), and it will call the right lvm command to fix it? For example, if a PV containing a side of a 2 way mirrored volume is lost, what are the various actions that it will do (I noticed an lvreduce --removemissing, but my fear is that this command will also remove unmirrored volumes)?

Thanks!

Bye!

Simone Gotti.

--
Email.it, the professional e-mail, gratis per te:
http://www.email.it/f

Sponsor:
Problemi di Liquidità? Con Logos Finanziaria 30.000 € in 24 ore a
dipendenti e lavoratori autonomi con rimborsi fino a 120 mesi, clicca qui
*
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=2911&d=18-4

Quoted 6 lines

_______________________________________________
linux-lvm mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

Quoted 5 lines

_______________________________________________
linux-lvm mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
Telefonare all'estero risparmiando fino all'80%? Con Email.it Phone Card puoi, clicca e scopri tutti i vantaggi Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=2683&d=20-4

#3 lore raw

thanks for the patch.

I've only compiled some of the kernels, since I'm still putting the finishing touches on the patches. I'll try compiling all of them on a spare machine and fix any problems, then repost the patches.

brassow

Quoted 162 lines · Simone Gotti
Quoted 3 lines · Jonathan E Brassow

Kernel patches are located at http://www.brassow.com/mirroring. Last updated on April 3rd... I'll be updating those either today or tomorrow as well.

Thanks,
I tried the version of the 3 April and the new version of yesterday. I applied the patches in the order provided by the "series" file, used the standard fedora .config, but on both the versions I got this compilation error:

CC [M] drivers/md/dm-log.o
drivers/md/dm-log.c: In function �dm_register_dirty_log_type�:
drivers/md/dm-log.c:19: error: �_lock� undeclared (first use in this function)
drivers/md/dm-log.c:19: error: (Each undeclared identifier is reported only once
drivers/md/dm-log.c:19: error: for each function it appears in.) drivers/md/dm-log.c: In function �dm_unregister_dirty_log_type�:
drivers/md/dm-log.c:29: error: �_lock� undeclared (first use in this function)
drivers/md/dm-log.c: In function �get_type�:
drivers/md/dm-log.c:46: error: �_lock� undeclared (first use in this function)
drivers/md/dm-log.c: In function �put_type�:
drivers/md/dm-log.c:64: error: �_lock� undeclared (first use in this function)
make[2]: *** [drivers/md/dm-log.o] Error 1
make[1]: *** [drivers/md] Error 2
make: *** [drivers] Error 2

I am missing something?

Quoted 2 lines

There are some kernel patches that you need. Either way, the DSO shouldn't be crashing, so I'll look into that.

My temporary workaround was to check the number of returned arguments:

Index: dmeventd_mirror.c
===================================================================
RCS file: /cvs/lvm2/LVM2/dmeventd/mirror/dmeventd_mirror.c,v retrieving revision 1.5
diff -u -b -B -p -r1.5 dmeventd_mirror.c

dmeventd_mirror.c 20 Apr 2006 20:42:18 -0000 renamed from dmeventd_mirror.c 6 Feb 2006 19:34:45 -0000 1.5 +8 -1
@@ -39,6 +39,7 @@ static int _get_mirror_event(char *param
 {
        int i, rtn = ME_INSYNC;
        int max_args = 30;  /* should support at least 8-way mirrors */
+       int output_args;
        char *args[max_args];
        char *dev_status_str;
        char *log_status_str;
@@ -46,11 +47,17 @@ static int _get_mirror_event(char *param
        char *p;
        int log_argc, num_devs, num_failures=0;
 
-       if (max_args <= split_words(params, max_args, args)) {
+       output_args = split_words(params, max_args, args);
+       if (max_args <= output_args) {
                syslog(LOG_ERR, "Unable to split mirror parameters: Arg
 list too long");
                return -E2BIG;
        }
 
+       if (output_args <= 4) {
+               syslog(LOG_ERR, "Mirror status not provided by the
 kernel module");
+               return ME_IGNORE;
+       };
+
        /*

* Unused: 0 409600 mirror
* Used : 2 253:4 253:5 400/400 1 AA 3 cluster 253:3 A

Thanks!

Bye!

Quoted 58 lines

brassow

Quoted 46 lines · Simone Gotti

Hi All,

I was curious to try the experimental dmeventd + the mirror checking dso.
Everytime I created a mirror volume, dmeventd was crashing when the mirror finished the syncing process.
I tracked it down to _get_mirror_event that is expeting a return string from the status ioctl to dm like "2 253:4 253:5 400/400 1 AA 3 cluster 253:3 A" but it gets only "2 253:4 253:5 400/400" so it SIGSEGV when is going to read over the number of returned arguments in the "args" array from split_words.

The same string is returned by "dmsetup status".

Do these informations aren't already implemented in the ioctl, or are there some patches around that I can try? (my kernel is the latest available in fedora rawhide: 2.6.16-1.2131_FC6)

Another question, If I understand it correctly, the mirror library for dmeventd will look to possible problems (eg the lost of a PV containing a mirror), and it will call the right lvm command to fix it? For example, if a PV containing a side of a 2 way mirrored volume is lost, what are the various actions that it will do (I noticed an lvreduce --removemissing, but my fear is that this command will also remove unmirrored volumes)?

Thanks!

Bye!

Simone Gotti.

--
Email.it, the professional e-mail, gratis per te:
http://www.email.it/f

Sponsor:
Problemi di Liquidit�? Con Logos Finanziaria 30.000 � in 24 ore a
dipendenti e lavoratori autonomi con rimborsi fino a 120 mesi, clicca qui
*
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=2911&d=18-4

Quoted 6 lines

_______________________________________________
linux-lvm mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

Quoted 5 lines

_______________________________________________
linux-lvm mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

--
Email.it, the professional e-mail, gratis per te:
http://www.email.it/f

Sponsor:
Telefonare all'estero risparmiando fino all'80%? Con Email.it Phone
Card puoi, clicca e scopri tutti i vantaggi
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=2683&d=20-4

_______________________________________________
linux-lvm mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/