Sun Enterprise Servers 3X00, 4X00, 5X00, 6X00 With "ecache" Entries in "/etc/system" May Hang During cpio(1) or ufsdump(1M) |
|
| Category : | Availability |
| Release Phase : | Resolved |
| Product : | Sun Enterprise 3000 Server Sun Enterprise 3500 Server Sun Enterprise 4000 Server Sun Enterprise 4500 Server Sun Enterprise 5500 Server Sun Enterprise 6500 Server
|
| Bug Id : | 4732826
|
| Date of Resolved Release : | 31-MAR-2004
|
Impact
In certain hardware configurations, Sun Enterprise Servers 3X00, 4X00, 5X00, 6X00 with "ecache" entries in the "/etc/system"file may experience SCSI errors while running the cpio(1) or ufsdump(1M) commands, eventually leading to an unresponsive (hung) system.
Contributing Factors
This issue can occur in the following configurations:
-
Sun Enterprise Servers (3X00/4X00/5X00/6X00) with Disk Card Option 501-5584 ("18.2 GB Wide SCSI Disk Card with 2x 9.1GB Disk Drives") containing a SCSI disk drive with part number (P/N) "DNES30917 SUN9.0G"
Only systems with "ecache" entries in the "/etc/system" file are affected by this issue, for example:
set ecache_scrub_enable=1
set ecache_scan_rate=1000
set ecache_calls_a_sec=100
Note: To determine if a system contains a "DNES30917 SUN9.0G" drive, do either one of the following:
-
while the system is at the OpenBoot PROM, run the "probe-scsi" command, or
-
from a command shell, start the format(1M) command as a root user, or
-
examine the "/opt/SUNWexplorer/bin/diskinfo" file generated by the "explorer" command
Symptoms
Should the described issue occur, messages similar to the following are logged to the "/var/adm/message" file:
Mar 18 10:08:28 tester unix: WARNING: /sbus@3,0/SUNW,fas@3,8800000/sd@b,0 (sd25):
Mar 18 10:08:28 tester SCSI transport failed: reason 'reset': retrying command
Mar 18 12:34:15 tester unix: NOTICE: alloc: /altroot: file system full
Mar 18 12:34:22 tester last message repeated 1 time
Workaround
To work around the described issue, do not use the cpio(1) or ufsdump(1M) commands or don't use the configuration as described above in the Contributing Factors section. Consider using tar(1) or dd(1M) instead.
Resolution
The final resolution is the Workaround provided above.
Modification HistoryDate: 31-MAR-2004
-
State: Resolved
-
Updated Relief/Workaround and Resolution sections
AttachmentsThis solution has no attachment