Important correction: it monitors the health of the first primary NFS (or otherwise “distributed and mounted”) filesystem. If you don’t use NFS als (main) primary storage, it’s safe to disable that reboot. If you know your NFS has “issues” from time to time, and have controls around that, yes, you can disable that…
Frank > On 15 Sep 2015, at 09:08, Abhinandan Prateek > <abhinandan.prat...@shapeblue.com> wrote: > > The heartbeat script monitors the health of the primary storage by using a > timestamp that is written to each primary store. > In case the primary storage is unreachable it reboots the XenServer in order > to protect the virtual machines from corruption. > >> On 14-Sep-2015, at 8:48 pm, Vadim Kimlaychuk <va...@kickcloud.net> wrote: >> >> Remi, >> >> I will definitely enable HA when find who is rebooting the host. I known >> circumstances when it happens and I know that it is storage-related. >> Hardware health is monitored by SNMP and there were no problems with >> temperature, CPU, RAM or HDD ranges. In case of HW failure I should >> theoretically have kernel panic or crash dumps. But there is none. Will >> experiment a bit. >> >> Thank you, >> >> Vadim. >> >> On 2015-09-14 17:35, Remi Bergsma wrote: >> >>> Hi Vadim, >>> It can also be XenHA but I remember you already said it is off. Did you >>> check the hardware health? >>> I'd recommend turning on XenHA as otherwise in case of a failure you will >>> not have an automatic recovery. >>> Regards, >>> Remi >>> On 14/09/15 15:09, "Vadim Kimlaychuk" <va...@kickcloud.net> wrote: >>> Remi, >>> I have analyzed script xenheartbeat.sh and it seems it is useless, >>> because relies on file /opt/cloud/bin/heartbeat that has 0 length. It is >>> not set-up during installation and there is no such a step in >>> documentation for setting it up. Logically admin must run >>> "setup_heartbeat_file.sh" to make heartbeat work. If this file is 0 >>> length then script checks nothing and log this message every minute: >>> Sep 14 04:43:53 xcp1 heartbeat: Problem with heartbeat, no iSCSI or NFS >>> mount defined in /opt/cloud/bin/heartbeat! >>> That means it can't reboot host, because it doesn't check >>> anything. Isn't it ? >>> Is there any other script that may reboot host if when there is a >>> problem with storage? >>> Vadim. >>> On 2015-09-14 15:40, Remi Bergsma wrote: >>> Hi Vadim, >>> This does indeed reboot a box, once storage fails: >>> echo b > /proc/sysrq-trigger >>> Removing it doesn't make sense, as there are serious issues once you >>> hit this code. I'd recommend making sure the storage is reliable. >>> Regards, Remi >>> On 14/09/15 08:13, "Vadim Kimlaychuk" <va...@kickcloud.net> wrote: >>> Remi, >>> I have analyzed situation and found that storage may cause problem >>> with host reboot as you wrote before in this thread. Reason for that -- >>> we do offline backups from NFS server at that time when hosts fail. >>> Basically we copy all files in primary and secondary storage offsite. >>> This process starts precisely at 00:00 and somewhere around 00:10 - >>> 00:40 XenServer host starts to reboot. >>> Reading old threads I have found that >>> /opt/cloud/bin/xenheartbeat.sh may do this job. Particularly last lines >>> at my xenheartbeat.sh are: >>> ------------------------- >>> /usr/bin/logger -t heartbeat "Problem with $hb: not reachable for >>> $(($(date +%s) - $lastdate)) seconds, rebooting system!" >>> echo b > /proc/sysrq-trigger >>> ------------------------- >>> The only "unclear" moment is -- I don't have such line in my logs. >>> May this command "echo b > /proc/sysrq-trigger" prevent from writing to >>> syslog file? Documentation says that it does reboot immediately without >>> synchronizing FS. It seems there is no other place that may do it, but >>> still I am not 100% sure. >>> Vadim. >>> On 2015-09-13 18:26, Vadim Kimlaychuk wrote: >>> Remi, >>> Thank you for hint. At least one problem is identified: >>> [root@xcp1 ~]# xe pool-list params=all | grep -E >>> "ha-enabled|ha-config" >>> ha-enabled ( RO): false >>> ha-configuration ( RO): >>> Where should I look for storage errors? Host? Management server? I have >>> checked /var/log/messages and there were only regular messages, no >>> "fence" or "reboot" commands. >>> I have dedicated NFS server that should be accessible all the time (at >>> least NIC interfaces are bonded in master-slave mode). Server is used >>> for both primary and secondary storage. >>> Thanks, >>> Vadim. >>> On 2015-09-13 14:38, Remi Bergsma wrote: >>> Hi Vadim, >>> Not sure what the problem is. Although I do know that when shared >>> storage is used, both CloudStack and XenServer will fence (reboot) the >>> box to prevent corruption in case access to the network or the storage >>> is not possible. What storage do you use? >>> What does this return on a XenServer?: >>> xe pool-list params=all | grep -E "ha-enabled|ha-config" >>> HA should be on, or else a hypervisor crash will not recover properly. >>> If you search the logs for Fence or reboot, does anything come back? >>> The logs you mention are nothing to worry about. >>> Can you tell us in some more details what happens and how we can >>> reproduce it? >>> Regards, >>> Remi >>> -----Original Message----- >>> From: Vadim Kimlaychuk [mailto:va...@kickcloud.net] >>> Sent: zondag 13 september 2015 9:32 >>> To: users@cloudstack.apache.org >>> Cc: Remi Bergsma >>> Subject: Re: CS 4.5.2: all hosts reboot after 3 days at production >>> Hello Remi, >>> This issue has nothing to do with CS 4.5.2. We got host reboot after >>> precisely 1 week with previous version of CS (4.5.1). Previous version >>> has been working without restart for 106 days before. So it is not a >>> software issue. >>> What does really make me unhappy -- accidental host reboot made entire >>> cluster unusable. Cloudstack management server was up and running, >>> second cluster node was up and running all the time and VM were >>> transferred to the second host, but System VMs were not rebooted >>> properly by CS and half of the network was down. SSVM and CPVM were in >>> "disconnected" status. Client VMs were up, but couldn't connect to >>> storage, because VRs were offline. Entire mess. >>> I have used planned maintenance mode before and cluster worked just >>> perfect. We didn't have any single second downtime. But with accidental >>> reboot there is no use of clusterization. :( >>> Vadim. >>> On 2015-09-08 09:35, Vadim Kimlaychuk wrote: >>> Hello Remi, >>> First of all I don't have /var/log/xha.log file. I have examined >>> logs >>> in detail and haven't found any trace that heartbeat has failed. The >>> only serious problem I have found in management logs before restart is >>> repeating many times error: >>> ----------------------------------- >>> -------------- >>> 2015-09-06 00:47:21,591 DEBUG [c.c.a.m.AgentManagerImpl] >>> (RouterMonitor-1:ctx-2d67d422) Details from executing class >>> com.cloud.agent.api.NetworkUsageCommand: Exception: >>> java.lang.Exception >>> Message: vpc network usage plugin call failed >>> Stack: java.lang.Exception: vpc network usage plugin call failed at >>> com.cloud.hypervisor.xenserver.resource.XenServer56Resource.VPCNetwork >>> Usage(XenServer56Resource.java:172) >>> at >>> com.cloud.hypervisor.xenserver.resource.XenServer56Resource.execute(Xe >>> nServer56Resource.java:195) >>> at >>> com.cloud.hypervisor.xenserver.resource.XenServer56Resource.executeReq >>> uest(XenServer56Resource.java:62) >>> at >>> com.cloud.hypervisor.xenserver.resource.XenServer610Resource.executeRe >>> quest(XenServer610Resource.java:87) >>> at >>> com.cloud.hypervisor.xenserver.resource.XenServer620SP1Resource.execut >>> eRequest(XenServer620SP1Resource.java:65) >>> at >>> com.cloud.agent.manager.DirectAgentAttache$Task.runInContext(DirectAge >>> ntAttache.java:302) >>> ... >>> ----------------------------------- >>> -------------- >>> Just couple of seconds before XCP2 host restart: >>> ----------------------------------- >>> -------------- >>> 2015-09-06 00:48:27,884 DEBUG [c.c.a.m.DirectAgentAttache] >>> (DirectAgentCronJob-83:ctx-ff822baf) Ping from 2(xcp1) >>> 2015-09-06 00:48:27,884 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) Process host VM state report >>> from >>> ping process. host: 2 >>> 2015-09-06 00:48:27,904 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) Process VM state report. host: >>> 2, >>> number of records in report: 6 >>> 2015-09-06 00:48:27,904 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM state report. host: 2, vm id: >>> 85, power state: PowerOn >>> 2015-09-06 00:48:27,907 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM power state does not change, >>> skip DB writing. vm id: 85 >>> 2015-09-06 00:48:27,907 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM state report. host: 2, vm id: >>> 1, power state: PowerOn >>> 2015-09-06 00:48:27,910 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM power state does not change, >>> skip DB writing. vm id: 1 >>> 2015-09-06 00:48:27,910 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM state report. host: 2, vm id: >>> 2, power state: PowerOn >>> 2015-09-06 00:48:27,913 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM power state does not change, >>> skip DB writing. vm id: 2 >>> 2015-09-06 00:48:27,913 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM state report. host: 2, vm id: >>> 82, power state: PowerOn >>> 2015-09-06 00:48:27,916 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM power state does not change, >>> skip DB writing. vm id: 82 >>> 2015-09-06 00:48:27,916 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM state report. host: 2, vm id: >>> 94, power state: PowerOn >>> 2015-09-06 00:48:27,919 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM power state does not change, >>> skip DB writing. vm id: 94 >>> 2015-09-06 00:48:27,919 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM state report. host: 2, vm id: >>> 90, power state: PowerOn >>> 2015-09-06 00:48:27,922 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) VM power state does not change, >>> skip DB writing. vm id: 90 >>> 2015-09-06 00:48:27,928 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-83:ctx-ff822baf) Done with process of VM state >>> report. host: 2 >>> 2015-09-06 00:48:27,940 DEBUG [c.c.a.m.DirectAgentAttache] >>> (DirectAgentCronJob-154:ctx-2e8a5911) Ping from 1(xcp2) >>> 2015-09-06 00:48:27,940 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) Process host VM state report >>> from ping process. host: 1 >>> 2015-09-06 00:48:27,951 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) Process VM state report. host: >>> 1, number of records in report: 4 >>> 2015-09-06 00:48:27,951 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) VM state report. host: 1, vm >>> id: >>> 100, power state: PowerOn >>> 2015-09-06 00:48:27,954 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) VM power state does not change, >>> skip DB writing. vm id: 100 >>> 2015-09-06 00:48:27,954 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) VM state report. host: 1, vm >>> id: >>> 33, power state: PowerOn >>> 2015-09-06 00:48:27,957 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) VM power state does not change, >>> skip DB writing. vm id: 33 >>> 2015-09-06 00:48:27,957 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) VM state report. host: 1, vm >>> id: >>> 89, power state: PowerOn >>> 2015-09-06 00:48:27,960 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) VM power state does not change, >>> skip DB writing. vm id: 89 >>> 2015-09-06 00:48:27,961 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) VM state report. host: 1, vm >>> id: >>> 88, power state: PowerOn >>> 2015-09-06 00:48:27,963 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) VM power state does not change, >>> skip DB writing. vm id: 88 >>> 2015-09-06 00:48:27,968 DEBUG >>> [c.c.v.VirtualMachinePowerStateSyncImpl] >>> (DirectAgentCronJob-154:ctx-2e8a5911) Done with process of VM state >>> report. host: 1 >>> ----------------------------------- >>> ----------------- >>> Vadim. >>> On 2015-09-07 23:18, Remi Bergsma wrote: >>> Hi Vadim, >>> What kind of storage do you use? Can you show /var/log/xha.log (I >>> think that is the name) please? It could be xen-ha that fences the box >>> if the heartbeat cannot be written. >>> You suggest it is CloudStack. Did you see anything in the mgt logs? >>> Regards, Remi >>> Sent from my iPhone >>> On 07 Sep 2015, at 08:26, Vadim Kimlaychuk <va...@kickcloud.net> wrote: >>> Hello all, >>> I have experienced accidental cluster reboot 3 days after update to CS >>> 4.5.2. Cluster is XenServer 6.5 with SP1. Reboot has been started from >>> slave node and then - master. >>> Syslog on slave shows only this: >>> Sep 6 00:47:05 xcp2 last message repeated 3 times Sep 6 00:47:15 xcp2 >>> xenstored: D12 write data/meminfo_free 713732 Sep 6 00:47:15 xcp2 >>> xenstored: A1564203 w event /local/domain/12/data/meminfo_free >>> /local/domain/12/data/meminfo_free >>> Sep 6 00:47:15 xcp2 xenstored: D12 write data/updated Sun Sep 6 >>> 00:48:55 EEST 2015 >>> Sep 6 00:47:15 xcp2 xenstored: A6 w event >>> /local/domain/12/data/updated /local/domain/12/data/updated >>> Sep 6 >>> 00:47:15 xcp2 xenstored: A10 w event /local/domain/12/data/updated >>> /local/domain/12/data/updated Sep 6 00:47:26 xcp2 dhclient: >>> DHCPREQUEST on xenbr0 to 172.17.0.1 port >>> 67 (xid=0x304ae9dc) >>> Sep 6 00:47:27 xcp2 xapi: [ info|xcp2|462044 INET >>> 0.0.0.0:80|dispatch:host.call_plugin D:7593b578fada|taskhelper] task >>> host.call_plugin R:ddd3cc399f86 forwarded >>> (trackid=407f6adaa118a34f19eb1e29cd68a0e8) >>> Sep 6 00:47:36 xcp2 dhclient: DHCPREQUEST on xenbr0 to 172.17.0.1 port >>> 67 (xid=0x304ae9dc) >>> Sep 6 00:48:18 xcp2 last message repeated 4 times Sep 6 00:48:25 xcp2 >>> xenstored: D1 write data/meminfo_free 1740496 Sep 6 00:48:25 xcp2 >>> xenstored: A1564203 w event /local/domain/1/data/meminfo_free >>> /local/domain/1/data/meminfo_free Sep 6 00:48:25 xcp2 xenstored: >>> D1 >>> write data/updated Sat Sep 5 21:50:07 EEST 2015 Sep 6 00:48:25 xcp2 >>> xenstored: A6 w event /local/domain/1/data/updated >>> /local/domain/1/data/updated Sep 6 00:48:25 xcp2 xenstored: A10 w >>> event /local/domain/1/data/updated >>> /local/domain/1/data/updated Sep 6 >>> 00:48:26 xcp2 dhclient: DHCPREQUEST on xenbr0 to 172.17.0.1 port >>> 67 (xid=0x304ae9dc) >>> Sep 6 00:48:27 xcp2 xapi: [ info|xcp2|462044 INET >>> 0.0.0.0:80|dispatch:host.call_plugin D:f2c8987bc0ff|taskhelper] task >>> host.call_plugin R:b62d2d4f58eb forwarded >>> (trackid=e3d4ea00c96194830a7dbbfc35563a3c) >>> Sep 6 00:48:38 xcp2 dhclient: DHCPREQUEST on xenbr0 to 172.17.0.1 port >>> 67 (xid=0x304ae9dc) >>> Sep 06 00:48:48 xcp2 syslogd 1.4.1: restart. >>> Sep 6 00:48:48 xcp2 kernel: klogd 1.4.1, log source = /proc/kmsg >>> started. >>> Sep 6 00:48:48 xcp2 kernel: [ 0.000000] Initializing cgroup subsys >>> cpuset Sep 6 00:48:48 xcp2 kernel: [ 0.000000] Initializing cgroup >>> subsys cpu Sep 6 00:48:48 xcp2 kernel: [ 0.000000] Initializing cgroup >>> subsys cpuacct Sep 6 00:48:48 xcp2 kernel: [ 0.000000] Linux version >>> 3.10.0+2 >>> Can anyone help with diagnostics ? >>> Thank you, >>> Vadim. > > Find out more about ShapeBlue and our range of CloudStack related services > > IaaS Cloud Design & Build<http://shapeblue.com/iaas-cloud-design-and-build//> > CSForge – rapid IaaS deployment framework<http://shapeblue.com/csforge/> > CloudStack Consulting<http://shapeblue.com/cloudstack-consultancy/> > CloudStack Software > Engineering<http://shapeblue.com/cloudstack-software-engineering/> > CloudStack Infrastructure > Support<http://shapeblue.com/cloudstack-infrastructure-support/> > CloudStack Bootcamp Training > Courses<http://shapeblue.com/cloudstack-training/> > > This email and any attachments to it may be confidential and are intended > solely for the use of the individual to whom it is addressed. Any views or > opinions expressed are solely those of the author and do not necessarily > represent those of Shape Blue Ltd or related companies. If you are not the > intended recipient of this email, you must neither take any action based upon > its contents, nor copy or show it to anyone. Please contact the sender if you > believe you have received this email in error. Shape Blue Ltd is a company > incorporated in England & Wales. ShapeBlue Services India LLP is a company > incorporated in India and is operated under license from Shape Blue Ltd. > Shape Blue Brasil Consultoria Ltda is a company incorporated in Brasil and is > operated under license from Shape Blue Ltd. ShapeBlue SA Pty Ltd is a company > registered by The Republic of South Africa and is traded under license from > Shape Blue Ltd. ShapeBlue is a registered trademark.
