| To: | pcp <pcp@xxxxxxxxxxx> |
|---|---|
| Subject: | Looking for unusual /proc/cpuinfo files |
| From: | Nathan Scott <nathans@xxxxxxxxxx> |
| Date: | Thu, 13 Nov 2014 18:35:56 -0500 (EST) |
| Delivered-to: | pcp@xxxxxxxxxxx |
| In-reply-to: | <1909930829.14254818.1415921115325.JavaMail.zimbra@xxxxxxxxxx> |
| Reply-to: | Nathan Scott <nathans@xxxxxxxxxx> |
| Thread-index: | QFljzzee6xxF4VKmx0VW6CO3MY+6Fg== |
| Thread-topic: | Looking for unusual /proc/cpuinfo files |
Hi all, The PCP Linux kernel PMDA parses /proc/cpuinfo to extract some detailed metrics about the hardware. One problem we have seen is that each architecture port places different (and sometimes completely different!) contents in this file, and we just have to deal with that in userspace as best we can. This came up the other day on aarch64, where values for some of the hinv.cpu.* metrics are incorrect but it affects other ports too. I'd like to start a QA effort to maintain static copies of some contents of /proc/cpuinfo & verify that each new PCP release is able to successfully parse them. If you have unusual machines, problematic cpuinfo contents, large CPU count machines - that sort of thing - please send me a copy of your /proc/cpuinfo and I'll add it to the set we test going forward. Thanks! -- Nathan |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | pcp updates - mixed bag, some QA, some init script tweaks, Ken McDonell |
|---|---|
| Next by Date: | Re: [Suspected SPAM] [pcp] New website, Shirshendu Chakrabarti |
| Previous by Thread: | pcp updates - mixed bag, some QA, some init script tweaks, Ken McDonell |
| Next by Thread: | pcp updates: merges (kenj,michele,lberk), Nathan Scott |
| Indexes: | [Date] [Thread] [Top] [All Lists] |