Scribd
Upload
3K views

Islion Interview Questions

The document provides troubleshooting steps for common Isilon cluster issues including nodes in attention state, permission errors, NFS/SMB issues, NTP issues, battery failure, unresponsive nodes, and collecting latest node logs. The key steps are to run diagnostic commands like "isi stat", "isi event list", check log files, user permissions, and network settings. If issues persist, contact EMC support and provide diagnostic logs collected using the "isi_gather_info" command. Replacing failed hardware like batteries may also be needed in some cases.

Uploaded by

Himanshu Mishra
Copyright
© © All Rights Reserved
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
3K views

Islion Interview Questions

The document provides troubleshooting steps for common Isilon cluster issues including nodes in attention state, permission errors, NFS/SMB issues, NTP issues, battery failure, unresponsive nodes, and collecting latest node logs. The key steps are to run diagnostic commands like "isi stat", "isi event list", check log files, user permissions, and network settings. If issues persist, contact EMC support and provide diagnostic logs collected using the "isi_gather_info" command. Replacing failed hardware like batteries may also be needed in some cases.

Uploaded by

Himanshu Mishra
Copyright
© © All Rights Reserved
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
You are on page 1/ 2

Islion Question

1) what are the steps which you are going to perform whenever you encountered cl
uster nodes are in "Attention" state:
Reason:a) avscan job failure
b) dblog is full
Answer: first we will run " isi stat" command on node to check its status.if s
tatus is showing as Attention perform below steps
1)collect islion logs by command "isi_gather_info",we may be prompted for your
root password, and then the logs from each of the nodes in the cluster will be c
ompiled into one file and automatically uploaded to us.
2) by Analysing the islion logs, we found that The issue here is identified as
Avscan failure event as the another job kicked in and is running.
3)run command "isi event list" to check the status of all jobs running on that n
ode.Note down the instanceid of failed avscan job.
4)cancel the job by command "isi events cancel --instanceids 3.2163
5) run command "isi stat", now status of node should be "ok"
6) check the size of dblog, if it is full then we have to clear the logs.
2) A Unix client is getting a permission denied error
Answer: 1) checking verifying quorum, ensuring that the cluster has not split an
d that the user is connecting to a node which can write to /ifs.
Next check whether quotas are in use, and if so how much room there is on the ta
rget directory. Check to see if the user has already exceeded their quota limit,
and then verify the mount options that they are using.
2) check the user by issuing the "isi auth users view username " command,Make su
re the user is in one of the groups that has write permissions.
3) now repair the permissions assigned :
login to GUI go to filesharing>advanced>repair permission job>select job option
To run the Repair Permission job:
1.In the web administration interface click PROTOCOLS, then click ACLs, and then
click Repair Permissions Job. The Repair Permissions Job page displays.
2.For Repair task, click to select one of the permission job tasks; clone, conve
rt, inherit.
3) nfsv4 mounted client reports permission problems accessing resource on cluste
r
1) check that nfsv4 domain string matches cluster
a) if domain doesn't matches
* set the resource to default owner/group or typically nobody/no group
The resolution is to set the @Domain string on the cluster to match exactly the
@Domain string on the Linux clients /etc/idmap.conf file. On the cluster, you can
set the string in the web administration interface click PROTOCOLS, then click
UNIX Sharing (NFS), and then click NFS Settings page.
4) how will you trouble shoot SMB issues if encountered:
1) run command isi smb session list
2)isi smb openfiles list
3)logs are stored in path /var/log/lwiod.log
default log level is error, we need to chnage it to debug
change log level:
isi smb log-level -- set=debug
5) how will you resolve ntp issues:
1)use the AD domain controller to set cluster time

2) to remove the ntp servers ,use the command:


isi_for_arrays -s "ntpq -n -p"
we can remove the NTP servers from the list and manually set the cluster s time
to match that of a domain controller.
we can set the cluster time from the web administration interface.
From the command line: # ntpq -c clocklist | grep timecode # isi_for_array s ntpq n
p
6) what will you do in case of battery failure
A failed battery condition will put the node into read-only mode and disable fur
ther use of NVRAM.
The NVRAM battery is a critical component of an EMC Isilon node to insure data i
ntegrity and provides write protection. The two potential issues are an unseated
or unconnected battery tray; and a battery failure. The follow these troublesho
oting steps to correct the issue
1) check battery tray and Confirm the battery tray has not been removed or is un
seated in the node.
Re-seat the battery tray to try to restore connectivity.
2) Run # isi batterystatus to see the current state of the NVRAM batteries and c
harging system.
3) If none of these possible solutions resolves the issue, contact EMC Isilon Su
pport, gather logs from the web administration interface. To do this, expand the
Help menu, click Diagnostics, under the Gather Info tab, and click Start Gather
4) ask emc to replace faulted battery asap.
7) Node is not responding issue
Ping that node from another node
Try to establish an SSH connection to the unresponsive node using an IP address a
ssigned to one of the Ethernet ports.
Try to establish an SSH connection to the unresponsive node through the switch se
rial port using an IP address assigned to the InfiniBand port attached to that s
witch.
Try to access the node from the serial port on the back of the cluster If the nod
e is still not responsive, perform a hard reboot of the node by turning it off a
nd then powering it back on.
During the boot sequence, the node performs a self test that could result in mor
e information about the status of the node. If the node is still unresponsive, w
e will log a call with emc.
8) is it possible to collect only latest logs of nodes
yes it is possible,we need to run command "isi_gather _info -- incremental"

You might also like