mke2fs slowdown

[rgreeott]

I have a Red Hat 7.2 (build 2.4.7-10smp) install with a Qlogic Qla2300 fibre channel adapter (driver version 4.28) and an attached IBM/Mylex DF4000 RAID enclosure. I have created several LUN's of 50 and 100GB. When mke2fs is called to make the filesystems, it seems to take an extra-ordinary amount of time to finish. Typically I can create a filesystem of 50GB in approximately 1 minute. This is taking on order of 10-20 minutes to complete per file system. I got a trace ( from a fibre channel analyzer) of mkfs running and then stalling or slowing down. It appears that the host just stops sending commands for extended periods of time. When ther DF4000 does get commands, it executes them in a timely manner. I looked for a reason for the host slowing down, such as us returning bad SCSI status or Link errors but I can't see anything. Just for grins I loaded mkfs 1.27 and it ran further without stalling or stuttering, but still hit the slowdown before completion.

Has anyone seen or heard of an issue like this before?

Thanks for any feedback....
Rick

[Stix]

Hell, that's easy... Everybody knows that Qlogic HBA's and Mylex RAID enclosures blow. You should have went with Ci Design products ;D ;D

Later my brotha.....

P.S. Welcome to the board

[pbharris]

hello,
any chance you are doing a surface scan while this is happening?

[rgreeott]

hello,
any chance you are doing a surface scan while this is happening?

No... No surface scans going on but there are other systems attached to the RAID device through a switch. However they don't have access to the LUN's in questions.

The FC trace I took shows the host (linux box) just stops sending commands for extended periods of time.