controlling tape device by udev rules (alias)

Question

hi,For any of you still using tape on Linux MA’s, I was wondering if there’s a way to force the Expert Storage Configuration to use the alias (friendly names) created by udev rules for those devices and the changer?Setting up udev rules isn’t that hard and I know RH says that WWNN’s used “by-id” are automagic, but the names are never human intuitive. Plus we’ve seen some issues whereby rebooting the MA is a crapshoot that means sometimes the drives all work and mount media fine, and sometimes it’s no dice. I could never see the changes noted in the events and so created udev rules to try and keep things consistent, but the field can’t be edited in expert storage config under any circumstances, it just finds the ../scsi-blah devices and not any of the aliases.

kkarthikk · Accepted Answer

Hi ​@downhillCommvault’s Expert Storage Configuration for Linux Media Agents does not support forcing device detection to use custom udev aliases (also known as friendly names) for tape drives or changers.The detection process relies exclusively on standard OS-reported SCSI device paths such as:/dev/sg*/dev/st*/dev/tape/by-id/*The device path field in Expert Storage Configuration is not editable. It auto-populates based on information provided by the operating system and kernel.While udev rules can create friendly symlinks for devices, Commvault does not provide a mechanism to manually select or enforce these aliases within the configuration.In certain environments—especially when using WWID or /dev/tape/by-id naming—users may experience inconsistencies after reboot, such as:Devices not being detectedDrives mounting under different pathsYou cannot force Commvault to use udev-friendly names for tape devices in Expert Storage Configuration. Device detection is restricted to standard SCSI paths as reported by the OS.Please let me know in case of further queries.Regards,Karthik

downhill · Answer

Hi ​@kkarthikkthanks for the clarification. That is exactly why I went down this path - having intermittent access issues to drives post-reboot. I understand that udev is better in the newer kernels but we never had these sorts of issues with Veritas (ever). The SN’s or WWID’s aren’t changing, but since adding the used rules it appears to me that inconsistent behavior has mostly subsided even if Commvault isn’t using those paths (the tape/by-id). I don't understand why it would be that hard to allow the expert admin to pin the logical drives to the preferred by-id’s - can’t this be made configurable?And if the answer is no - then the solution is - random hit or miss following reboots followed by either “issues” with jobs dying due to a single stream failing the entire job (but not until the very end of “running” streams)?? If no one else has experienced this I’d sure like to know where it comes from. We’ve been working with RH support as well on this but no solid explanation or fix yet.thanks

Sign up

Login to the community

Scanning file for viruses.

This file cannot be downloaded