[multipath] meaning of "no_path_retry nn"

[multipath] meaning of "no_path_retry nn"

[Diedrich Ehlerding]

I would appreciate an explanation of "no_path_retry nn" parameter in 
multipath.conf. Probably I misunderstood it as a number of seconds. 

I am setting up  a software mirror (mdadm) on top of an EMC Clariion 
configuration for desaster protection. The Clariions are recognized by 
multipath properly; it seems to be set up with "no_path_retry 60" 
(according to multipath -t). I had no further settings in 
multipath.conf; this seems to be hardcoded into multipath. 

However, If I disconnect one Clariion, th mirror doesnt break - 
multipath detects that all paths are failed and tells me that it will 
now wait for 60 retries. These 60 retries seem to last very long; 
atfter 5 minutes, it didnt complete. 

Evertyhing works fine if I set up a multipath.conf containing the 
Clariion settings from multipath -t and modify them to 
"no_path_retry fail" and "features 0 fail_if_no_path".  But even if I 
just reduce the default settings to no_path_retry 2 the mirror doesnt 
break - the IOs seem to hang indefinitely (well, I lost patience after 
~5 minutes).

Probably I misunderstood "no_path_retry 60" - does that mean seconds, 
or minutes, or what else? Is anything except "fail" a useful setting in 
mirrored configurations? 

system is a Suse  with multipath-tools-0.4.7-34.43

Thanks 
Diedrich
-- 
Diedrich Ehlerding, Fujitsu Technology Solutions GmbH, R GE TIS N IC2 
Hildesheimer Str 25, D-30880 Laatzen
Fon +49 511 8489-1806, Fax -251806, Mobil +49 173 2464758
Firmenangaben: http://de.ts.fujitsu.com/imprint.html

RE: [multipath] meaning of "no_path_retry nn"

[Moger, Babu]

This is how the time is calculated. 
Total queing time =  nn* polling_interval.

For example 
	polling_interval        10
	no_path_retry 		5

This will queue-up the i/os for 50 seconds. You should see i/o error after 50 seconds in case of all path failure.

However, this may not work in all the distros. We have seen problems with this feature. Looks like upstream tool has fixed this problem.
 
Thanks
Babu Moger 

> -----Original Message-----
> From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com]
> On Behalf Of Diedrich Ehlerding
> Sent: Friday, August 28, 2009 8:16 AM
> To: dm-devel@redhat.com
> Subject: [dm-devel] [multipath] meaning of "no_path_retry nn"
> 
> I would appreciate an explanation of "no_path_retry nn" parameter in
> multipath.conf. Probably I misunderstood it as a number of seconds.
> 
> I am setting up  a software mirror (mdadm) on top of an EMC Clariion
> configuration for desaster protection. The Clariions are recognized by
> multipath properly; it seems to be set up with "no_path_retry 60"
> (according to multipath -t). I had no further settings in
> multipath.conf; this seems to be hardcoded into multipath.
> 
> However, If I disconnect one Clariion, th mirror doesnt break -
> multipath detects that all paths are failed and tells me that it will
> now wait for 60 retries. These 60 retries seem to last very long;
> atfter 5 minutes, it didnt complete.
> 
> Evertyhing works fine if I set up a multipath.conf containing the
> Clariion settings from multipath -t and modify them to
> "no_path_retry fail" and "features 0 fail_if_no_path".  But even if I
> just reduce the default settings to no_path_retry 2 the mirror doesnt
> break - the IOs seem to hang indefinitely (well, I lost patience after
> ~5 minutes).
> 
> Probably I misunderstood "no_path_retry 60" - does that mean seconds,
> or minutes, or what else? Is anything except "fail" a useful setting in
> mirrored configurations?
> 
> system is a Suse  with multipath-tools-0.4.7-34.43
> 
> Thanks
> Diedrich
> --
> Diedrich Ehlerding, Fujitsu Technology Solutions GmbH, R GE TIS N IC2
> Hildesheimer Str 25, D-30880 Laatzen
> Fon +49 511 8489-1806, Fax -251806, Mobil +49 173 2464758
> Firmenangaben: http://de.ts.fujitsu.com/imprint.html
> 
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel

Re: [multipath] meaning of "no_path_retry nn"

[Hannes Reinecke]

Diedrich Ehlerding wrote:
> I would appreciate an explanation of "no_path_retry nn" parameter in 
> multipath.conf. Probably I misunderstood it as a number of seconds. 
> 
> I am setting up  a software mirror (mdadm) on top of an EMC Clariion 
> configuration for desaster protection. The Clariions are recognized by 
> multipath properly; it seems to be set up with "no_path_retry 60" 
> (according to multipath -t). I had no further settings in 
> multipath.conf; this seems to be hardcoded into multipath. 
> 
> However, If I disconnect one Clariion, th mirror doesnt break - 
> multipath detects that all paths are failed and tells me that it will 
> now wait for 60 retries. These 60 retries seem to last very long; 
> atfter 5 minutes, it didnt complete. 
> 
> Evertyhing works fine if I set up a multipath.conf containing the 
> Clariion settings from multipath -t and modify them to 
> "no_path_retry fail" and "features 0 fail_if_no_path".  But even if I 
> just reduce the default settings to no_path_retry 2 the mirror doesnt 
> break - the IOs seem to hang indefinitely (well, I lost patience after 
> ~5 minutes).
> 
> Probably I misunderstood "no_path_retry 60" - does that mean seconds, 
> or minutes, or what else? Is anything except "fail" a useful setting in 
> mirrored configurations? 
> 
> system is a Suse  with multipath-tools-0.4.7-34.43
> 
Please note that there is a bug in the multipath-tools on suse which
prevents the no_path_retry option from working correctly.

Please install the latest maintenance update.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		      zSeries & Storage
hare@suse.de			      +49 911 74053 688
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: Markus Rex, HRB 16746 (AG Nürnberg)