NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.
To learn about what's going on, see About the Archive Site.
Can you check if Qpid is bottle-necked? qpid-stat -q -S msg -L 10 On 10/24/2013 12:57 PM, Tyle, Kevin R wrote:
Load is back to the neighborhood of 25-30 again. Attached is output from iostat and sar. I've also run htop as suggested. It typically shows cpu usage varying from little more than 5% on most of the 8 processors to occasional periods when all eight are between 70-85%. I'm not sure what to look for in terms of slow disk writes/reads being the culprit. The fourth internal disk in the system is where all the EDEX/LDM-related data is filed. Thanks, Kevin _____________________________________________ Kevin Tyle, Systems Administrator Dept. of Atmospheric& Environmental Sciences University at Albany Earth Science 235, 1400 Washington Avenue Albany, NY 12222 Email: ktyle@xxxxxxxxxx Phone: 518-442-4578 _____________________________________________ -----Original Message----- From: daryl herzmann [mailto:akrherz@xxxxxxxxxxx] Sent: Wednesday, October 23, 2013 2:01 PM To: Tyle, Kevin R Cc: awips2-users@xxxxxxxxxxxxxxxx Subject: Re: [awips2-users] CPU load on EDEX/LDM server Kevin, It would be good to interogate system monitoring tools like 'sar' to see what it is logging for CPU vs IO wait percentages. In general, high CPU load is typically tied to processes / threads waiting for IO to occur. If your disk is unable to keep up with raw IO or intensive metadata operations, that could be the trouble. Running 'top' will also show you if your processes are chewing actual CPU or the load average is artificially high with IO wait being to blame. daryl On Wed, 23 Oct 2013, Tyle, Kevin R wrote:A follow-up: the high CPU load was almost certainly linked to the ldm server that was running. Once it was shut down, load dropped to close to 0. Upon restart, load jumped back up to the vicinity of 20.00. I then shut down the LDM and purged the data directories (which were very very large since no scouring had been done). Restarting the LDM shows low system load so far. I will enable LDM scouring and keep an eye on things. _____________________________________________ Kevin Tyle, Systems Administrator Dept. of Atmospheric& Environmental Sciences University at Albany Earth Science 235, 1400 Washington Avenue Albany, NY 12222 Email: ktyle@xxxxxxxxxx<mailto:ktyle@xxxxxxxxxx> Phone: 518-442-4578 _____________________________________________ From: awips2-users-bounces@xxxxxxxxxxxxxxxx [mailto:awips2-users-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Tyle, Kevin R Sent: Tuesday, October 22, 2013 11:00 AM To: awips2-users@xxxxxxxxxxxxxxxx Subject: [awips2-users] CPU load on EDEX/LDM server Hi, I am running an EDEX server, also running LDM, using the 13.2 version that Michael has made available to us. It is on a CentOS5 platform, 32-bit, 16 CPU, 12 GB RAM. I am not running CAVE on this machine. I regularly see an average load around 20.00, which seems excessive. Lots of java, httpd, postmaster processes owned by user awips. Does this load seem excessive? What do others see? If this load indeed seems excessive, any ideas to troubleshoot? Thanks, Kevin _____________________________________________ Kevin Tyle, Systems Administrator Dept. of Atmospheric& Environmental Sciences University at Albany Earth Science 235, 1400 Washington Avenue Albany, NY 12222 Email: ktyle@xxxxxxxxxx<mailto:ktyle@xxxxxxxxxx> Phone: 518-442-4578 _____________________________________________-- /** * Daryl Herzmann * Assistant Scientist -- Iowa Environmental Mesonet * http://mesonet.agron.iastate.edu */_______________________________________________ awips2-users mailing list awips2-users@xxxxxxxxxxxxxxxx For list information, to unsubscribe, or change your membership options, visit: http://www.unidata.ucar.edu/mailing_lists/
awips2-users
archives: