Hi Jim. Most certainly XRC can slow your I/O down and it may be too much for linux. There is a parm somewhere on the GDPS side called write pacing that may be in play here. That keeps you from writing too fast and bogging down the XRC network so much that RPO is missed. Are u sharing datamovers with VM and MVS? Do you see a difference in your io response time numbers than you did before?
Talk to your folks in charge of replication and definitely open a prob with IBM if they can't help. If you do get 256 of those your linux file system will go read only and that's not fun at all. Marcy. Sent from my BlackBerry. ----- Original Message ----- From: [email protected] [mailto:[email protected]] Sent: Tuesday, February 04, 2014 06:02 PM Central Standard Time To: [email protected] <[email protected]> Subject: [LINUX-390] DASD Interrupt Errors Related to XRC I/O Delays? Hello List, We have a SLES11 SP4 Linux server running under z/VM 6.2 on a zEC12 that's generating the following message repeatedly during periods of peak I/O activity: kernel: dasd_erp(3990): 0.0.01bn: dasd_3990_erp_action_4: first time retry where n=1, 6 or 9 All 3 of these minidisks reside on the same Linux formatted VM volume (ECKD). I discovered that these messages are generated by the IBM DASD Error Recovery Procedure (ERP). After looking at the Linux code surrounding this ERP error (action 4), it appears that it can potentially retry up to 256 times, but I only see this "first time retry" message occurrence - but repeatedly, which suggests to me that these delays are of a fairly short duration (the code indicates the use of a 20 second timer between retries after the first retry). I also discovered that XRC DASD mirroring was made active just last week for the volume that these minidisks reside on. We have been running XRC DASD mirroring for a long time between MVS systems, but just started using it with our Linux production volumes. So, I suspect that these messages are the result of I/O interrupts caused by delays due to channel extenders, which are a part of the XRC configuration that associates this servers DASD volume with a counterpart at another data center over 700 miles away. Also, this server runs DataStage and gets pretty busy during peak periods due to extract, transform and load (ETL) processing. So, I have a couple of questions related to these ERP errors: 1. Have others seen these ERP messages related to XRC activity during peak I/O periods or can confirm that is what this is likely related to? 2. If it is related, then are these messages typical? If typical, is there a "tuning knob" that can adjust the amount of delay more appropriately to allow for this XRC delay during peak periods? Any help is greatly appreciated! Thanks in advance, Jim Moling IT Specialist, z/VM & Linux on z Mainframe Services Branch Division of Platform Services Information and Security Services Bureau of the Fiscal Service Department of the Treasury [email protected] 202-874-9566 ----------------------------------------- This E-mail and its attachments (if any) are intended solely for the use of the addressee(s) and may contain sensitive but unclassified information. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, or use of the information contained herein (including any reliance thereon) is strictly prohibited. If you have received this E-mail in error, please notify the sender immediately and destroy the E-mail and any attachments. ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For more information on Linux on System z, visit http://wiki.linuxvm.org/ ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For more information on Linux on System z, visit http://wiki.linuxvm.org/
