| Title: | DIGITAL UNIX (FORMERLY KNOWN AS DEC OSF/1) |
| Notice: | Welcome to the Digital UNIX Conference |
| Moderator: | SMURF::DENHAM |
| Created: | Thu Mar 16 1995 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 10068 |
| Total number of notes: | 35879 |
Hi,
an our A8400 has gone in a singular situation:
it seem stop all process (it is became not accessible from network)
and at console appear this message in loop:
tlaser: Can't allocate error log buffer
tlaser: Can't allocate error log buffer
tlaser: Can't allocate error log buffer
tlaser: Can't allocate error log buffer
tlaser: Can't allocate error log buffer
tlaser: Can't allocate error log buffer
Eh?
log buffer
tlaser: Can't allocate error log buffer
.....
.....
halted CPU0 <-we have digit a CTRL/P
halt code = 2
kernel stack not valid halt
tlaser: Can't alloca
Eh?
log buffer
tlaser: Can't allocate error log
Eh?
ffer
tlaser: Can't allocate error log buffer
halted CPU0 <-we have digit a CTRL/P
halt code =2
kernel stack not valid halt
PC = fffffc00004e5edc
CPU0 restarting
.......
tlaser: Can't alloca
Eh?
log buffer
tlaser: Can't allocate error log
Eh?
ffer
tlaser: Can't allocate error log buffer
P00>>>S <-we have digit a CTRL/P and
an involontary s
00067AC8 ! RET R31,(R26) dH
at this time other CTRL/P cannot help.
No errorlog entry can be found, no message in /var/adm/messages and
/var/adm/syslog.dated/.../syslog.log etc etc , has be registred
at time of the hang.
What do you think about ?
How investigate about ?
Alphaserver 8400 console V4.8-6 SROMV3.1
4 Cpu's kn7cc-ab
4 GB Ram ms7cc-ab
Operating System: Dec unix 3.2g
Thanks for your help
Gabriele
[Posted by WWW Notes gateway]
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 9984.1 | slow system? binlogd not running? | GIDDAY::STRAUSS | talking through my binoculars | Thu May 29 1997 19:55 | 18 |
Hi Gabriele
I think the error log buffer is circular. When it is full, it starts
writing at the beginning again.
If the system is running slowly, and the old messages have not been
written from the buffer to the error log file, then there is no room to
write new messages into the buffer. In this case, you might see the
"Can't allocate error log buffer" message.
Or perhaps binlogd is not running at all?
Orr perhaps something is trying to write an exceptionally large message
to the buffer?
Hope this helps
leon
| |||||
| 9984.2 | Other hypothesis... | NNTPD::"[email protected]" | Gabriele Mattacchione | Fri May 30 1997 06:52 | 41 |
Thanks for your help, I can add that the binlogd is running on the system and that we had restart the hsz40 that contain the /usr because when at reboot of the system it attempt to mount, the filesystem respond with following messages: ...... INIT: SINGLE-USER MODE -: /usr_opt_svil/cauni/scripts/envusr: not found p_isr 0x 6, p_cdc_addrhig 0xf00f, p_cfap* p_semaphore 0x 0, p_ddc_config 0x 0, p_dfap* p_nvram 0x 0, p_ddc_control 0x 0 p_cdc_config 0x 0, p_ddc_status 0x 8 s_pic 0x 2, s_uflags 0x 124, s_q* .................. .................. isp2: Firmware revision 2.10 (loaded from driver) cam_logger: CAM_ERROR packet cam_logger: bus 2 isp_cam_bus_reset_tmo SCSI Bus Reset performed the machine didn't exit from this error unless we restart the hsz40 that contain the /usr disks. Is not possible that the problem was happened from scsi interface and the HSZ40 device ? What behaviour can have Dec Unix if it lose the /usr ? Gabriele [Posted by WWW Notes gateway] | |||||
| 9984.3 | re -1 | KYOSS1::GREEN | Fri May 30 1997 17:11 | 4 | |
-1
The boot error can probably be fixed by installing isp1020.o patch.
you should install all patches + isp1020.
| |||||
| 9984.4 | another hypothesis | GIDDAY::STRAUSS | talking through my binoculars | Sun Jun 01 1997 19:30 | 16 |
Hi Gabriele
Another hypothesis.
If you were unable to mount /usr, and if /var is a link to /usr/var,
then the errlog log file would be unavailable.
And if the file was unavailable, binlogd would be unable to flush its
buffers. So the buffers fill up, and you get the error message.
It's just a guess.
Run # ls -ld /var or # file /var to see if it's a link to /usr/var
Hope this helps
leon
| |||||