summaryrefslogtreecommitdiffstats
path: root/docs/troubleshooting.txt
blob: fd5715022219fd4983ad2a88240293b5ffaf4ce4 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
The services has to be running
------------------------------
  Etcd:
    - etcd 

  Node:
    - origin-node
 
  Master nodes:
    - origin-master-api
    - origin-master-controllers
    - origin-master is not running

  Required Services:
    - lvm2-lvmetad.socket 
    - lvm2-lvmetad.service
    - docker                                    - it may happen that service is alive according to systemd, but does not respond ('docker ps' timeouts)
    - NetworkManager
    - firewalld
    - dnsmasq
    - openvswitch
 
  Extra Services:
    - ssh
    - ntp
    - openvpn
    - ganesha (on master nodes, optional)

Pods has to be running
----------------------
  Kubernetes System
    - kube-service-catalog/apiserver
    - kube-service-catalog/controller-manager
  
  OpenShift Main Services
    - default/docker-registry
    - default/registry-console
    - default/router (3 replicas)
    - openshift-template-service-broker/api-server (daemonset, on all nodes)

  OpenShift Secondary Services
    - openshift-ansible-service-broker/asb
    - openshift-ansible-service-broker/asb-etcd

  GlusterFS
     - glusterfs-storage (daemonset, on all storage nodes)
     - glusterblock-storage-provisioner-dc
     - heketi-storage

  Metrics (openshift-infra):
    - hawkular-cassandra
    - hawkular-metrics
    - heapster
    

Debugging
=========
 - Ensure system consistency as explained in 'consistency.txt' (incomplete)
 - Check current pod logs and possibly logs for last failed instance
        oc logs <pod name> --tail=100 [-p]                  - dc/name or ds/name as well
 - Verify initialization steps (check if all volumes are mounted)
        oc describe <pod name>
 - Security (SCC) problems are visible if replica controller is queried
        oc -n adei get  rc/mysql-1 -o yaml
 - It worth looking the pod environment
        oc env po <pod name> --list
 - It worth connecting running container with 'rsh' session and see running processes,
 internal logs, etc. The 'debug' session will start a new instance of the pod.
 - If try looking if corresponding pv/pvc are bound. Check logs for pv.
    * Even if 'pvc' is bound. The 'pv' may have problems with its backend.
    * Check logs here: /var/lib/origin/plugins/kubernetes.io/glusterfs/
 - Another frequent problems is failing 'postStart' hook. Or 'livenessProbe'. As it
 immediately crashes it is not possible to connect. Remedies are:
    * Set larger initial delay to check the probe.
    * Try to remove hook and execute it using 'rsh'/'debug'
 - Determine node running the pod and check the host logs in '/var/log/messages'
    * Particularly logs of 'origin-master-controllers' are of interest
 - Check which docker images are actually downloaded on the node
        docker images

network
=======
 - There is a NetworkManager script which should adjust /etc/resolv.conf to use local dnsmasq server.
 This is based on  '/etc/NetworkManager/dispatcher.d/99-origin-dns.sh' which does not play well 
 if OpenShift is running on non-default network interface. I provided a patched version, but it
 worth verifying 
    * that nameserver is pointing to the host itself (but not localhost, this is important
    to allow running pods to use it)
    * that correct upstream nameservers are listed in '/etc/dnsmasq.d/origin-upstream-dns.conf'
    * that correct upstream nameservers are listed in '/etc/origin/node/resolv.conf'
    * In some cases, it was necessary to restart dnsmasq (but it could be also for different reasons)
 If script misbehaves, it is possible to call it manually like that
    DEVICE_IFACE="eth1" ./99-origin-dns.sh eth1 up


etcd (and general operability)
====
 - Few of this sevices may seem running accroding to 'systemctl', but actually misbehave. Then, it 
 may be needed to restart them manually. I have noticed it with 
    * lvm2-lvmetad.socket       (pvscan will complain on problems)
    * node-origin
    * glusterd in container     (just kill the misbehaving pod, it will be recreated)
    * etcd               but BEWARE of too entusiastic restarting:
 - However, restarting etcd many times is BAD as it may trigger a severe problem with 
 'kube-service-catalog/apiserver'. The bug description is here
        https://github.com/kubernetes/kubernetes/issues/47131
 - Due to problem mentioned above, all 'oc' queries are very slow. There is not proper
 solution suggested. But killing the 'kube-service-catalog/apiserver' helps for a while.
 The pod is restarted and response times are back in order.
    * Another way to see this problem is quering 'healthz' service which would tell that
    there is too many clients and, please, retry later.
        curl -k https://apiserver.kube-service-catalog.svc/healthz

 - On node crash, the etcd database may get corrupted. 
    * There is no easy fix. Backup/restore is not working.
    * Easiest option is to remove the failed etcd from the cluster.
        etcdctl3 --endpoints="192.168.213.1:2379" member list
        etcdctl3 --endpoints="192.168.213.1:2379" member remove <hexid>
    * Add it to [new_etcd] section in inventory and run openshift-etcd to scale-up etcd cluster.
 
 - There is a helth check provided by the cluster
    curl -k https://apiserver.kube-service-catalog.svc/healthz
 it may complain about etcd problems. It seems triggered by OpenShift upgrade. The real cause and
 remedy is unclear, but the installation is mostly working. Discussion is in docs/upgrade.txt
 
 - There is also a different etcd which is integral part of the ansible service broker: 
 'openshift-ansible-service-broker/asb-etcd'. If investigated with 'oc logs' it complains 
 on:
        2018-03-07 20:54:48.791735 I | embed: rejected connection from "127.0.0.1:43066" (error "tls: failed to verify client's certificate: x509: certificate signed by unknown authority", ServerName "")
        WARNING: 2018/03/07 20:54:48 Failed to dial 0.0.0.0:2379: connection error: desc = "transport: authentication handshake failed: remote error: tls: bad certificate"; please retry.
 Nevertheless, it seems working without much trouble. The error message seems caused by
 certificate verification code which introduced in etcd 3.2. There are multiple bug repports on
 the issue.
 
pods (failed pods, rogue namespaces, etc...)
====
 - OpenShift has numerous problems with clean-up resources after the pods. The problems are more likely to happen on the 
 heavily loaded systems: cpu, io, interrputs, etc.
    * This may be indicated in the logs with various errors reporting inability to stop containers/processes, free network
    and storage resources. A few examples (not complete)
        dockerd-current: time="2019-09-30T18:46:12.298297013Z" level=warning msg="container kill failed because of 'container not found' or 'no such process': Cannot kill container 00a456097fcf8d70a0461f05813e5a1f547446dd10b3b43ebc1f0bb09e841d1b: rpc error: code = 2 desc = no such process"
        origin-node: W0930 18:46:11.286634    2497 util.go:87] Warning: Unmount skipped because path does not exist: /var/lib/origin/openshift.local.volumes/pods/6aecbed1-e3b2-11e9-bbd6-0cc47adef0e6/volumes/kubernetes.io~glusterfs/adei-tmp
        Error syncing pod 1ed138cd-e2fc-11e9-bbd6-0cc47adef0e6 ("adei-smartgrid-maintain-1569790800-pcmdp_adei(1ed138cd-e2fc-11e9-bbd6-0cc47adef0e6)"), skipping: failed to "CreatePodSandbox" for "adei-smartgrid-maintain-1569790800-pcmdp_adei(1ed138cd-e2fc-11e9-bbd6-0cc47adef0e6)" with CreatePodSandboxError: "CreatePodSandbox for pod \"adei-smartgrid-maintain-1569790800-pcmdp_adei(1ed138cd-e2fc-11e9-bbd6-0cc47adef0e6)\" failed: rpc error: code = 2 desc = NetworkPlugin cni failed to set up pod \"adei-smartgrid-maintain-1569790800-pcmdp_adei\" network: CNI request failed with status 400: 'failed to Statfs \"/proc/28826/ns/net\": no such file or directory\n'"
    * A more severe form is then PLEG (POD Lifecycle Event Generator) errors are reported:
        origin-node: I0925 07:52:00.422291   93115 kubelet.go:1796] skipping pod synchronization - [PLEG is not healthy: pleg was last seen active 3m0.448988393s ago; threshold is 3m0s]
    This indicates a severe delays in communication with docker daemon (can be checked with 'docker info') and may result in node marked
    temporarily NotReady causing 'pod' eviction. As pod eviction causes extensive load on the other nodes (which may also be affected of the
    same problem), the initial single-node issue may render all cluster unusable.
    * With mass evictions, the things could get even worse causing faults in etcd communication. This is reported like:
         etcd: lost the TCP streaming connection with peer 2696c5f68f35c672 (stream MsgApp v2 reader)
    * Apart from overloaded nodes (max cpu%, io, interrupts), PLEG issues can be caused by 
        1. Excessive amount of resident docker images on the node (see bellow)
        2. This can cause and will be further amplified by the spurious interfaces on OpenVSwich (see bellow)
        x. Nuanced issues between kubelet, docker,   logging, networking and so on, with remediation of the issue sometimes being brutal (restarting all nodes etc, depending on the case).
            https://github.com/kubernetes/kubernetes/issues/45419#issuecomment-496818225
    * The problem is not bound to CronJobs, but having regular scheduled jobs make it presence significantly more visible. 
    Furthermore, CronJobs especially scheduling fat containers, like ADEI, significantly add to the I/O load on the system 
    and may cause more severe form.

 - After a while, the 'pods' schedulling may get more-and-more sluggish, in general or if assigned to a specific node.
  * The docker images are accumulating on the nodes over time. After a threshold it will start adding the latency to the 
  operation of docker daemon, slow down the pod scheduling (on the affected nodes), and may cause other sever side effects. 
  The problems will start appearing at around 500-1000 images accumulated at a specific node. With 2000-3000, it will get 
  severe and almost unusable (3-5 minutes to start a pod). So, eventually the unused images should be cleaned
    oc adm prune images --keep-tag-revisions=3 --keep-younger-than=60m --confirm
  or alternatively per-node:
    docker rmi $(docker images --filter "dangling=true" -q --no-trunc)
  * Some images could be orphanned by OpenShift infrastructure (there was not a major number of orphaned images on KaaS yet).
  OpenShift supports 'hard' prunning to handle such images.
    https://docs.openshift.com/container-platform/3.7/admin_guide/pruning_resources.html
  * Even afterwards, a significant number of images may stay resident. There is two inter-related problems:
    1. Docker infrastructure relies on the intermediate images. Consequently, very long Dockerfiles will create a LOT of images. 
    2. OpenShift keeps history of 'rc' which may refence several versions of old docker images. This will be not cleaned by the
    described approach. Furthermore, stopped containers lost by OpenShift infrastructure (see above) also prevent clean-up of
    the images
  Currenly, a dozen KDB pods produce about 200-300 images. In some cases, optimization of dockerfiles and, afterwards, a trough
  cleanup of old images may become necessity. The intermediate images can be found with 'docker images -a' (all images with
  <none> as repository and the name), but there is no easy way to find pod populating them. One, but not very convinient is the following
  project (press F5 on startup): https://github.com/TomasTomecek/sen

 - In a more sever4 form, the 'pods' scheduling may fail all together on one (or more) of the nodes. After a long waiting,
 the 'oc logs' will report timeout. The 'oc describe' reports 'failed to create pod sandbox'. This can be caused by failure 
 to clean-up after terminated pod properly. It causes rogue network interfaces to remain in OpenVSwitch fabric. 
  * This can be determined by errors reported using 'ovs-vsctl show' or present in the log '/var/log/openvswitch/ovs-vswitchd.log' 
    which may quickly grow over 100MB quickly. 
        could not open network device vethb9de241f (No such device)
  * The work-around is to delete rogue interfaces with 
        ovs-vsctl del-port br0 <iface>
    More info:
        ovs-ofctl -O OpenFlow13 show br0
        ovs-ofctl -O OpenFlow13 dump-flows br0
    This does not solve the problem, however. The new interfaces will get abandoned by OpenShift.
  * The issue is discussed here:
        https://bugzilla.redhat.com/show_bug.cgi?id=1518684
        https://bugzilla.redhat.com/show_bug.cgi?id=1518912

 - After crashes / upgrades some pods may end up in 'Error' state. This is quite often happen to
    * kube-service-catalog/controller-manager
    * openshift-template-service-broker/api-server
 Normally, they should be deleted. Then, OpenShift will auto-restart pods and they likely will run without problems.
    for name in  $(oc get pods -n openshift-template-service-broker | grep Error | awk '{ print $1 }' ); do oc -n openshift-template-service-broker delete po $name; done
    for name in  $(oc get pods -n kube-service-catalog | grep Error | awk '{ print $1 }' ); do oc -n kube-service-catalog delete po $name; done 
 
 - Other pods will fail with 'ImagePullBackOff' after cluster crash. The problem is that ImageStreams populated by 'builds' will 
 not be recreated automatically. By default OpenShift docker registry is stored on ephemeral disks and is lost on crash. The build should be 
 re-executed manually.
        oc -n adei start-build adei

 - Furthermore, after long outtages the CronJobs will stop functioning. The reason can be found by analyzing '/var/log/messages' or specially
        systemctl status origin-master-controllers
  it will contain something like:
        'Cannot determine if <namespace>/<cronjob> needs to be started: Too many missed start time (> 100). Set or decrease .spec.startingDeadlineSeconds or check clock skew.'
  * The reason is that after 100 missed (or failed) launch periods it will stop trying to avoid excive load. The remedy is set 'startingDeadlineSeconds'
  which tells the system that if cronJob has failed to start in the allocated interval we stop trying until the next start period. Then, 100 is only 
  counted the specified period. I.e. we should set period bellow the 'launch period / 100'.
        https://github.com/kubernetes/kubernetes/issues/45825
  * The running CronJobs can be easily patched with
        oc -n adei patch cronjob/adei-autogen-update --patch '{ "spec": {"startingDeadlineSeconds": 120 }}'
  
 - Sometimes there is rogue namespaces in 'deleting' state. This is also hundreds of reasons, but mainly
    * Crash of both masters during population / destruction of OpenShift resources
    * Running of 'oc adm diagnostics'
  It is unclear how to remove them manually, but it seems if we run
    * OpenShift upgrade, the namespaces are gone (but there could be a bunch of new problems).
    * ... i don't know if install, etc. May cause the trouble...

 - There is also rogue pods (mainly due to some problems with unmounting lost storage) remaining "Deleting" state, etc. 
 There are two possible situations:
    * The containers are actually already terminated, but OpenShift is not aware of it for some reason.
    * The containers are actually still running, but OpenShift is not able to terminate them for some reason.
 It is relatively easy to find out which is the case:
    * Finding the host running the failed pod with 'oc get pods -o wide'
    * Checking if associated containers are still running on the host with 'docker ps'
 The first case it relatively easy to handle, - one can simply enforce pod removal with
        oc delete --grace-period=0 --force
 In the second case we need 
    * To actually stop containers before proceeding (enforcing will just leave them running forever). This can
    be done directly using 'docker' commands.
    * It also may be worth trying to clean associated resources. Check 'maintenace' documentation for details.

 - Permission problems will arise if non-KaaS namespace (using high range supplemental-group for GlusterFS mounts) is converted
 to KaaS (gid ranges within 1000 - 10,000 at the moment). The allowed gids should be configured in the namespace specification 
 and the pods should be allowed to access files. Possible errors:
    unable to create pods: pods "mongodb-2-" is forbidden: no providers available to validate pod request



Builds
======
 - After changing storage for integrated docker registry, it may refuse builds with HTTP error 500. It is necessary
 to run:
    oadm policy reconcile-cluster-roles


Storage
=======
 - The offline bricks can be brough back into the service with the follwoing command
    gluster volume start openshift force

 - Running a lot of pods may exhaust available storage. It worth checking if 
    * There is enough Docker storage for containers (lvm)
    * There is enough Heketi storage for dynamic volumes (lvm)
    * The root file system on nodes still has space for logs, etc. 
  Particularly there is a big problem for ansible-ran virtual machines. The system disk is stored
  under '/root/VirtualBox VMs' and is not cleaned/destroyed unlike second hard drive on 'vagrant
  destroy'. So, it should be cleaned manually.

 - Too many parallel mounts (above 500 per node) may cause systemd slow-down/crashes. It is indicated by
 the following messages in the log:
        E0926 09:29:50.744454   93115 mount_linux.go:172] Mount failed: exit status 1
        Output: Failed to start transient scope unit: Connection timed out
    * Solution is unclear, there are some suggestions to use 'setsid' in place of 'systemd-run' to do mounting,
    but not clear how. Discussion: https://github.com/kubernetes/kubernetes/issues/79194
    * Can we do some rate-limiting?
  
 - Problems with pvc's can be evaluated by running 
        oc  -n openshift-ansible-service-broker describe pvc etcd
   Furthermore it worth looking in the folder with volume logs. For each 'pv' it stores subdirectories
   with pods executed on this host which are mount this pod and holds the log for this pods.
        /var/lib/origin/plugins/kubernetes.io/glusterfs/

 - Heketi is problematic.
    * Worth checking if topology is fine and running.
        heketi-cli -s http://heketi-storage-glusterfs.openshift.suren.me --user admin --secret "$(oc get secret heketi-storage-admin-secret -n glusterfs -o jsonpath='{.data.key}' | base64 -d)"
 - Furthermore, the heketi gluster volumes may be started, but with multiple bricks offline. This can
 be checked with 
        gluster volume status <vol> detail
    * If not all bricks online, likely it is just enought to restart the volume
        gluster volume stop <vol>
        gluster volume start <vol>
    * This may break services depending on provisioned 'pv' like 'openshift-ansible-service-broker/asb-etcd'
    
 - If something gone wrong, heketi may end-up creating a bunch of new volumes, corrupt database, and crash
 refusing to start. Here is the recovery procedure.
    * Sometimes, it is still possible to start by setting 'HEKETI_IGNORE_STALE_OPERATIONS' environmental
    variable on the container.
        oc -n glusterfs env dc  heketi-storage -e HEKETI_IGNORE_STALE_OPERATIONS=true
    * Even if it works, it does not solve the main issue with corruption. It is necessary to start a 
    debugging pod for heketi (oc debug) export corrupted databased, fix it, and save back. Having
    database backup could save a lot of hussle to find that is amiss.
        heketi db export --dbfile heketi.db --jsonfile /tmp/q.json
        oc cp glusterfs/heketi-storage-3-jqlwm-debug:/tmp/q.json q.json
        cat q.json | python -m json.tool > q2.json
        ...... Fixing .....
        oc cp q2.json glusterfs/heketi-storage-3-jqlwm-debug:/tmp/q2.json 
        heketi db import --dbfile heketi2.db --jsonfile /tmp/q2.json
        cp heketi2.db /var/lib/heketi/heketi.db
    * If bunch of disks is created, there are still various left-overs. First, the Gluster volumes
    have to be cleaned. The idea is to compare 'vol_' prefixed volumes in Heketi and Gluster. And
    remove ones not present in heketi. There is the script in 'ands/scripts'.
    * There is LVM volumes left from Gluster (or even allocated, but not associated with Gluster for
    various failurs. so this clean-up is worth making independently). On each node we can easily find
    volumes created today
        lvdisplay -o name,time -S 'time since "2018-03-16"'
    or again we can compare lvm volumes which are used by Gluster bricks and which are not. The later
    ones should be cleaned up. Again there is the script.
     
MySQL
=====
 - MySQL may stop replicating from the master. There is some kind of deadlock in multi-threaded SLAVE SQL.
 This can be seen by exexuting (which should show a lot of slave threads waiting on coordinator to provide
 the load).
        SHOW PROCESSLIST;
 The remedy is to restart slave MySQL with 'slave_parallel_workers=0', give it a time to go, and then
 restart back in the standard multithreading mode. This can be achieved by editing 'statefulset/mysql-slave-0'
 and setting environmental vairable 'MYSQL_SLAVE_WORKERS' to 0 and, then, back to original value (16 currently).

- This could be not end of this. The execution of statments from the log could 'stuck' because of the some "bad"
  transaction. This can be detected by looking into the replication status on the slave database:
        SHOW SLAVE STATUS\G
 It may happen what while 'Retrieved_Gtid_Set' increments (binary log is successfully transfered from the master
 and contains new transactions), the 'Executed_Gtid_Set' staying constant, i.e. new transactions are not applied 
 to the slave server. Also, in this case 'Relay_Log_Pos' and 'Exec_Master_Log_Pos' will not advance. And the
 'Slave_SQL_Running_State' will likely stuck in the 'Reading event from the relay log' state. The solution is to 
 skip misbehaving transaction. Basically, we need to skip problematic transaction and to continue from the next 
 one. The following considers that GTID_MODE is 'on' as we have it on the current KaaS platform.
  * First, we need to find the next transaction. Likely it the just an increment of the 'Executed_Gtid_Set', i.e.
    if 'Executed_Gtid_Set=3beaee24-2e55-11e8-9612-0a580a80000c:1-953787306' on the slave, the next GTID would be:
    'Executed_Gtid_Set=3beaee24-2e55-11e8-9612-0a580a80000c:953787307' (yes "1-" also goes away). But this is not
    guaranteed. The more reliable way to find the misbehaving transaction is to look into the 'bin-log' on the master
    server (information is taken from SHOW SLAVE STATUS executed on the slave). In '/var/lib/mysql/data' run
        mysqlbinlog --start-position=<Exec_Master_Log_Pos> <Relay_Master_Log_File> | head -n 50
    Here you will find which transaction has likely caused the problem. Furhermore, there will be the line looking like
        SET @@SESSION.GTID_NEXT='4ab8feff-5272-11e8-9320-08002715584a:201840'
    This is the gtid of the next transaction.
 * So, the following commands should be executed on the slave MySQL server (see details, https://www.thegeekdiary.com/how-to-skip-a-transaction-on-mysql-replication-slave-when-gtids-are-enabled/)
        SLAVE STOP;
        SET @@SESSION.GTID_NEXT='<found_gtid_of_next_transaction>';
        BEGIN;
        COMMIT;
        SET GTID_NEXT='AUTOMATIC';
        SLAVE START;
 * It is also possible to review the stuck transaction on the slave mysql node. In the '/var/lib/mysql/data' run 
        mysqlbinlog --start-position=<Relay_Log_Pos> <Relay_Log_File>
    
Administration
==============
 - Some management tasks may require to login on ipekatrin* nodes. Thereafter, the password-less execution of
 'oc' may fail on master nodes complaining on invalid authentication token. To fix it, it is necessary to check
 /root/.kube/config and remove references on logged users keeping only 'system:admin/kaas-kit-edu:8443' alkso check
 listed contexts and current-context.
     
Performance
===========
 - To find if OpenShift restricts the usage of system resources, we can 'rsh' to container and check
 cgroup limits in sysfs
    /sys/fs/cgroup/cpuset/cpuset.cpus
    /sys/fs/cgroup/memory/memory.limit_in_bytes


Various
=======
 - IPMI may cause problems as well. Particularly, the mounted CDrom may start complaining. Easiest is
 just to remove it from the running system with
     echo 1 > /sys/block/sdd/device/delete