Custom Query (1359 matches)
Results (136 - 138 of 1359)
| Ticket | Resolution | Summary | Owner | Reporter |
|---|---|---|---|---|
| #145 | fixed | smartctl -l scterc,70,70 with RAID5 on ICH10R Onboard Raid | ||
| Description |
Hi! There is no way to send any SMART commands to an Raid array which is built by the Intel ICH10R onboard Raid controller. No issue when using the drives as single ones without raid. The Tool "Argus Monitor" is able to report SMART status values, even if the drives are in a raid array. Is there a possibility to get smartmontools working to send the scterc even if it is part of an array? Would be very very very great... Many thanks in advance |
|||
| #146 | fixed | smartctl does not work on arbitrary symlinks to devices | ||
| Description |
$ ln -s /dev/sda /tmp/foo $ smartctl /tmp/foo smartctl 5.40 2010-07-12 r3124 [x86_64-unknown-linux-gnu] (local build) Copyright (C) 2002-10 by Bruce Allen, http://smartmontools.sourceforge.net /tmp/foo: Unable to detect device type ... $ smartctl /dev/sda # this works fine $ smartctl -a /dev/block/8:0 # FAIL, as above $ smartctl /dev/disk/by-path/pci-0000\:00\:1f.2-scsi-0\:0\:0\:0 # this works fine, too Even if all the links point to /dev/sda, only some of them can be specified as arguments to smartctl - which does not make sence IMHO. I would expect any symbolic link (which may be nested quite deep) that finally resolves to a device should allow successful execution of smartctl. In my use case, I have quite a bunch of disks that are used for experiments and switched between machines regularily. Therefore I set up symlinks like /data/dev/disk1 -> /dev/disk/by-id/ABC-123456 with a fixed mapping of the disk$i names and prefer to access the devices via /data/dev/disk$i (and mount them to /data/disk$i) so I don't have to care which /dev/sdX name it currently has (and I don't accidentally access the wrong one - I don't care about the port where it is connected but about the actual physical disk). Using /dev/disk/by-id/ABC-123456 directly is quite cumbersome because they are quite long, look different from vendor to vendor and are not contiguous. UUIDs or something similar is not useful either because the disks get repartitioned and filesystems get recreated regularily. While sfdisk, mkfs, mount, hdparm, ... work in this setup, smartctl does not. Andreas |
|||
| #147 | fixed | smartctl logs results of old, and outdated selftest | ||
| Description |
Test #19 failed, but since then the sector is fixed (somehow, reallocated or read successfully). Test #2 now shows success. Smartctl is ignoring the new success status, and constantly logging the failure. (which propagates up the monitoring chain) Can smartctl be modified, as default or via an option, to use the last "Completed" instead of an old "Completed" ? (I've snipped a number of lines from the output below: START OF READ SMART DATA SECTIONSMART Self-test log structure revision number 1 Num Test_Description Status Remaining LifeTime(hours) LBA_of_first_error # 1 Extended offline Interrupted (host reset) 90% 22506 - # 2 Extended offline Completed without error 00% 22488 - #19 Extended offline Completed: read failure 90% 20054 824368247 #20 Extended offline Interrupted (host reset) 90% 19903 - Regards, John Jore |
|||
