Hey,
You got hit by a few things. What version of mogilefs are you running? 2.17?
device 7 is not in DB, nor is any file ( _* table) linked to this device.
device 7 was a spare hd (/dev/hdc1) with data on it ... i planned to
insert it into mogile after i inserted the data on it into mogilefs
(turned out to be a defect harddisk anyway)
(after assembling the computer it was hdc on the second comp in cluster
so it was meant to be dev7)
but back to the errors:
after i added dev7 as dead device to host 2 all worked fine and no
errors anymore.
This is due (IIRC) to an issue where the monitor process was actually
scanning for all devices available on a host, and passing that
information back into the mogile tracker without a related DB for said
devices.
another wired problem is that file_to_replicate had strange values.
it contained 90 rows with a nexttry value of 2147483647 (translates to
Tuesday, January 19th 2038, 3:14:07 (GMT)).
That's a "magic" value (ENDOFTIME) or end_of_time(). It means the fid
completely failed to replicate and needs user interaction to fix. I
admit there's a complete lack of interface around this "feature" of
mogilefs :) But they were likely related dev7 as you said.
device 7 is still a noshow (though visible in db as dead device):
[EMAIL PROTECTED]:~# mogadm --trackers=10.10.10.132:7001 device list
fsc1 [1]: alive
used(G) free(G) total(G)
dev1: alive 14.662 222.814 237.477
dev2: alive 17.594 257.479 275.072
dev3: alive 14.987 168.396 183.384
dev4: alive 41.199 142.185 183.384
fsc2 [2]: alive
used(G) free(G) total(G)
dev5: alive 20.813 237.761 258.574
dev6: alive 24.292 269.118 293.410
dev8: alive 23.549 269.861 293.410
Dead devices are no longer printed in mogadm output.
Guh, need to cut a release :)
-Dormando
