Not having any joy with replacing the broken glusterfs node.
What we did was:
1. Delete the broken gluster node from the cluster, and remove it from
the inventory file
2. Create a new node to replace it. Add it to the [new_nodes] section of
the inventory and run the playbooks/byo/openshift-node/scaleup.yml
playbook. At this stage it is not added to the [glusterfs] section of
the inventory. The node is now part of the cluster. Move it from the
[new_nodes] section of the inventory to the [nodes] section.
3. Add the new node to the [glusterfs] section of the inventory. At
this stage we have the 2 functioning gluster nodes with volumes
containing data, and one new node with unformatted volumes.
4. Edit the [OSEv3:vars] section and add these 3 properties:
openshift_storage_glusterfs_wipe = False
openshift_storage_glusterfs_is_missing = False
openshift_storage_glusterfs_heketi_is_missing = False
5. Run the playbooks/byo/openshift-glusterfs/config.yml playbook. This
fails with the following error:
TASK [openshift_storage_glusterfs : Load heketi topology]
************************************************************************************************************************************
Saturday 25 August 2018 12:04:37 +0000 (0:00:02.073) 0:26:39.480 *******
fatal: [orn-master.openstacklocal]: FAILED! => {"changed": true,
"cmd": ["oc",
"--config=/tmp/openshift-glusterfs-ansible-2Zl8Vv/admin.kubeconfig",
"rsh", "--namespace=glusterfs", "heketi-storage-2-jvfhn",
"heketi-cli", "-s", "http://localhost:8080";, "--user", "admin",
"--secret", "sjGJ1Gix0Nf9GEaXynTSngMwi6D/fHtEEWxyZCSlVY8=",
"topology", "load",
"--json=/tmp/openshift-glusterfs-ansible-2Zl8Vv/topology.json",
"2>&1"], "delta": "0:00:05.354851", "end": "2018-08-25
12:10:10.128102", "failed_when_result": true, "rc": 0, "start":
"2018-08-25 12:10:04.773251", "stderr": "", "stderr_lines": [],
"stdout": "\tFound node orn-gluster-storage-001.openstacklocal on
cluster de03021c7b9d5f6a99d403a7a369d3e1\n\t\tFound device
/dev/vdb\n\t\tFound device /dev/vdc\n\t\tFound device
/dev/vdd\n\tFound node orn-gluster-storage-002.openstacklocal on
cluster de03021c7b9d5f6a99d403a7a369d3e1\n\t\tFound device
/dev/vdb\n\t\tFound device /dev/vdc\n\t\tFound device
/dev/vdd\n\tCreating node orn-gluster-storage-003.openstacklocal ...
Unable to create node: Unable to execute command on
glusterfs-storage-k7lp4: peer probe: failed: Probe returned with
Transport endpoint is not connected", "stdout_lines": ["\tFound node
orn-gluster-storage-001.openstacklocal on cluster
de03021c7b9d5f6a99d403a7a369d3e1", "\t\tFound device /dev/vdb",
"\t\tFound device /dev/vdc", "\t\tFound device /dev/vdd", "\tFound
node orn-gluster-storage-002.openstacklocal on cluster
de03021c7b9d5f6a99d403a7a369d3e1", "\t\tFound device /dev/vdb",
"\t\tFound device /dev/vdc", "\t\tFound device /dev/vdd", "\tCreating
node orn-gluster-storage-003.openstacklocal ... Unable to create node:
Unable to execute command on glusterfs-storage-k7lp4: peer probe:
failed: Probe returned with Transport endpoint is not connected"]}
Similarly, if you oc rsh to the heketi pod and run the heketi-cli you
get a similar error:
heketi-cli node add --zone=1 --cluster=$CLUSTER_ID
--management-host-name=orn-gluster-storage-003.openstacklocal
--storage-host-name=10.0.0.26
Error: Unable to execute command on glusterfs-storage-k7lp4: peer
probe: failed: Probe returned with Transport endpoint is not connected
Any thoughts how to repair this?
On 24/08/18 13:39, Walters, Todd wrote:
Tim,
Try deleting all the pods, the glusterfs pods and the heketi pod. Do it one at
a time. I’ve had this work for me where the pods came back up and heketi was ok.
Also can try restarting glusterfs glusterd in the pod term on each pod. That’s
worked for me to get out of heketi db issues.
Other than that I don’t have any other ideas. I’ve not found good information
on how to resolve or troubleshoot issues like this.
Thanks,
Todd
On 8/24/18, 4:37 AM, "Tim Dudgeon" <tdudgeon...@gmail.com> wrote:
Todd,
Thanks for that. Seems on the lines that I need.
The problem though is that I have an additional problem of the heketi
pod not starting because of a messed up database configuration.
These two problems happened independently, but on the same OpenShift
environment.
This means I'm unable to run the heketi-cli until that is fixed.
I'm not sure if I can modify the heketi database configuration as
described in the troubleshooting guide [1] so that it only knows about
the two good gluster nodes, and then add back the third one?
Any thoughts?
Tim
[1]
https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fheketi%2Fheketi%2Fblob%2Fmaster%2Fdocs%2Ftroubleshooting.md&data=01%7C01%7CTodd_Walters%40unigroup.com%7Cee469430dd3e43c27c2508d609a5284e%7C259bdc2f86d3477b8cb34eee64289142%7C1&sdata=uzBTRZV59mpOQfdWWO%2FMFmwIZO8NIu6cpPp%2Bjz%2BD4Yc%3D&reserved=0
On 23/08/18 17:14, Walters, Todd wrote:
> Tim,
>
> I have had this issue with 3 node cluster. I created a new node with new
devices, ran scaleup and ran gluster playbook with some changes, then ran
heketi-cli commands to add new node and remove old node.
>
> For your other question, I’ve restarted all glusterfs pods and hekeit
pod and resolved that issue before. I guess you can restart glusterd in each pod
too?
>
> Here’s doc I wrote on node replacement. I’m not sure if this is proper
procedure, but it works, and I wasn’t able to find any decent solution in the docs.
>
> # ----- Replacing a Failed Node ---- #
>
> Disable Node to simulate failure
> Get node id with heketi-cli node list or topology info
>
> heketi-cli node disable fb344a2ea889c7e25a772e747eeeec2a -s http://localhost:8080
--user admin --secret "$HEKETI_CLI_KEY"
> Node fb344a2ea889c7e25a772e747eeeec2a is now offline
>
> Stop Node in AWS Console
> Scale up another node (4) for Gluster via Terraform
> Run scaleup_node.yml playbook
>
> Add New Node and Device
>
> heketi-cli node add --zone=1 --cluster=441248c1b2f032a93aca4a4e03648b28
--management-host-name=ip-new-node.ec2.internal --storage-host-name=newnodeIP -s
http://localhost:8080 --user admin --secret "$HEKETI_CLI_KEY"
> heketi-cli device add --name /dev/xvdc --node 8973b41d8a4e437bd8b36d7df1a93f06 -s
http://localhost:8080 --user admin --secret "$HEKETI_CLI_KEY"
>
>
> Run deploy_gluster playbook, with the following changes in OSEv3
>
> - openshift_storage_glusterfs_wipe: False
> - openshift_storage_glusterfs_is_missing: False
> - openshift_storage_glusterfs_heketi_is_missing: False
>
> Verify topology
> rsh into heketi pod
> run heketi-exports (file i created with export commands)
> get old and new node info (id)
>
> Remove Node
>
> sh-4.4# heketi-cli node remove fb344a2ea889c7e25a772e747eeeec2a -s
http://localhost:8080 --user admin --secret "$HEKETI_CLI_KEY"
> Node fb344a2ea889c7e25a772e747eeeec2a is now removed
>
>
> Remove All Devices (check the topology)
>
> sh-4.4# heketi-cli device delete ea85942eaec73cb666c4e3dcec8b3702 -s
http://localhost:8080 --user admin --secret "$HEKETI_CLI_KEY"
> Device ea85942eaec73cb666c4e3dcec8b3702 deleted
>
>
> Delete the Node
>
> sh-4.4# heketi-cli node delete fb344a2ea889c7e25a772e747eeeec2a -s
http://localhost:8080 --user admin --secret "$HEKETI_CLI_KEY"
> Node fb344a2ea889c7e25a772e747eeeec2a deleted
>
>
> Verify New Topology
>
> $ heketi-cli topology info
> make sure new node and device is listed.
>
>
> Thanks,
>
> Todd
>
> # -----------------------
>
> Check any existing pvc is still accessible.
> Today's Topics:
> 2. Replacing failed gluster node (Tim Dudgeon)
>
>
> ------------------------------
>
> Message: 2
> Date: Thu, 23 Aug 2018 15:40:29 +0100
> From: Tim Dudgeon <tdudgeon...@gmail.com>
> To: users <users@lists.openshift.redhat.com>
> Subject: Replacing failed gluster node
> Message-ID: <b3999ba8-d4bd-16e1-6135-ca5d9ed76...@gmail.com>
> Content-Type: text/plain; charset=utf-8; format=flowed
>
> I have a 3 node containerised glusterfs setup, and one of the nodes
has
> just died.
> I believe I can recover the disks that were used for the gluster
storage.
> What is the best approach to replacing that node with a new one?
> Can I just create a new node with empty disks mounted and use the
> scaleup.yml playbook and [new_nodes] section, or should I be
creating a
> node that re-uses the existing drives?
>
> Tim
>
>
>
> ------------------------------
>
> _______________________________________________
> users mailing list
> users@lists.openshift.redhat.com
>
https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Flists.openshift.redhat.com%2Fopenshiftmm%2Flistinfo%2Fusers&data=01%7C01%7CTodd_Walters%40unigroup.com%7Cee469430dd3e43c27c2508d609a5284e%7C259bdc2f86d3477b8cb34eee64289142%7C1&sdata=ut6aI0zzqt4mGBImtX0zTdgm1LNKFVv5ksE1eYjQG40%3D&reserved=0
>
>
> End of users Digest, Vol 73, Issue 44
> *************************************
>
>
>
> ########################################################################
> The information contained in this message, and any attachments thereto,
> is intended solely for the use of the addressee(s) and may contain
> confidential and/or privileged material. Any review, retransmission,
> dissemination, copying, or other use of the transmitted information is
> prohibited. If you received this in error, please contact the sender
> and delete the material from any computer. UNIGROUP.COM
> ########################################################################
>
########################################################################
The information contained in this message, and any attachments thereto,
is intended solely for the use of the addressee(s) and may contain
confidential and/or privileged material. Any review, retransmission,
dissemination, copying, or other use of the transmitted information is
prohibited. If you received this in error, please contact the sender
and delete the material from any computer. UNIGROUP.COM
########################################################################
_______________________________________________
users mailing list
users@lists.openshift.redhat.com
http://lists.openshift.redhat.com/openshiftmm/listinfo/users
