tester | I've been playing (quite a bit) with suspend to disk and resume. Something funny has happened though. During these "re-cyclings," at some point my drives went from being /dev/sda and /dev/sdb to where it is now /dev/sda and /dev/sdd. | 07:06 |
---|---|---|
tester | I understand that can happen sometimes from a boot. | 07:06 |
tester | These drives, on a test box, were /dev/sda5, /dev/sdb5, and /dev/md7p4. | 07:08 |
tester | I have all of my drives in /etc/fstab set up with UUIDs rather than the /dev/mapper or /dev/sd* paths. | 07:08 |
tester | I have not mucked with these since successfully setting up the suspend/resume business. | 07:09 |
tester | But now, I am seeing /dev/sdb5 as one of my swap drives! | 07:09 |
tester | the system is humming along nicely, but I find this odd... and probably, kinda... wrong? | 07:09 |
tester | but /dev/sda5 does not appear in swapon -s | 07:10 |
tester | I disabled the two drives whose uuids match to /dev/sda5 and /dev/sdb5 | 07:10 |
tester | oh, and one other thing: The swap partition I am using for hibernate/resume is not either /dev/sd?; I am using that 3rd drive, /dev/md7p4 | 07:12 |
tester | If I swapon -a, I see the /dev/md7p4 swap but I also see the /dev/sdb5 swap (which should not be there, right?) | 07:13 |
tester | So I think what is happening, in part at least, is that what used to be /dev/sdb5 is now /dev/sdd5. And somehow, when the system resumed, after one of my testings, the /dev/sdb drive became /dev/sdd. | 07:16 |
tester | Now, this IS a test box, and I do put it through some beatings. But in this case, I'd thin the linux kernel would not even think about a device /dev/sdb* if it has assigned the same drive /dev/sdd now. | 07:18 |
tester | s/thin/think/ | 07:19 |
tester | And I admit I may have done something to screw things up somehow. | 07:20 |
tester | (I'm still a rookie with some of this stuff, though I think I have md raid and even cryptsetup down) | 07:20 |
tester | "down"--poor choice of words there. I mean, I understand them now. | 07:21 |
tester | and LVM also, which I have been using for some time, although I do not use the many flavors of LV's besides the sort of vanilla type. | 07:21 |
ted-ious | tester: Have you tried rescanning all the scsi busses to see what it does? | 07:40 |
ted-ious | Maybe that needs to be done whenever you resume. | 07:40 |
tester | scan fu please? | 07:44 |
tester | echo "- - -" > /sys/class/scsi_host/host0/scan | 07:46 |
tester | maybe? | 07:46 |
tester | but you mean during the boot process? | 07:46 |
tester | if it needs to be done, then I would hope the utilities support hibernate and suspend / resume would handle that already. | 07:47 |
tester | this is actually the only weirdness I've noticed since starting to muck with the hibernate stuff | 07:47 |
YKaelig | Hello | 15:04 |
al1r4d | Hello | 15:12 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!