Displaying Retired Memory Pages with /usr/bin/mdb -k
Hello everyone,
Got 15 "Page <hex>,<hex> removed from service" in my /var/adm/message* files :(
I would like to see if they are REALLY retired. There's a useful sundoc on the subject, using "mdb"
http://sunsolve.sun.com/search/document.do?assetkey=1-9-76833-1&searchclaus e=memory%20errors%20page%20retired
Tells me to do
# /usr/bin/mdb -k
> retired_ppages::walk page
but I get an error message:-
mdb: failed to perform walk: unknown walk name
Has anyone done anything like this? I've used the "kstat -s retired_pages" command to. It tell me I've got 15 retired pages. I take it the memory DIMMS will need replacing, or can I just get away will re-seating them?
The server is a E3500. Solaris 8 kernel patch 117350-08. The E3500 has 2 system boards. Each board has 2 CPU's (400Mhz) and 2Mb RAM. Thus 4 CPUs & 4Mb RAM.
Thanks for any help.
Patrick/
[960 byte] By [
patrick275] at [2007-11-25 23:35:02]
# 1
This is pretty good at telling you when to replace DIMMs: http://sunsolve.sun.com/search/document.do?assetkey=1-9-70361-1
You may want to call your Solution Centre and find out where to download the cediag tool mentioned in :
http://sunsolve.sun.com/search/document.do?assetkey=1-9-79928-1
# 2
Thanks for the reply :)
I will follow up on the "cediag" program. Sun says it's downloadable from sunsolv, but I haven't seen it yet. Regarding replacement of DIMMS: From one of your links it says:-
NOTE: Prior to Solaris 10 OS, retired pages are returned to service whenever
a system is rebooted, and will be re-retired if Solaris encounters CEs from
the same page again. DO NOT replace any DIMMs unless the page retirement
limit is reached for that DIMM
Will I be notifed then the "page retirement limit" has been reached in /var/adm/messages? I have 15 entries so far saying "Page 0x0000000,<hex> removed from service". But nothing denoting a limit has been reached...so I guess I should replace any DIMMS yet ;)
BTW: I haven't rebooted...and I don't want to unless really necessary.
Patrick/
