Note: You can run the GPFS defrag command while the file systems are
mounted. However, for better results, unmount the GPFS file system
before performing the defragmentation operation.
2. If there is no space in fragments or if the mmdefragfs command does not free
up space, add disks (NSDs) to the file system to create space.
a. Add disks to the file system.
Note: If free space exists in the mdiskgroup then you can modify the file
system by editing it in the GUI or simply running the command: mkdisk
fileSystem size mdiskgroup
For example:
[root@kd01gln.mgmt002st001 ~]# mkdisk gpfs0 10GB 0
(1/4) Creating Storage System volumes
(2/4) Scanning for new devices
(3/4) Creating NSDs
(4/4) Adding disks to filesystem
Successfully created disk
b. If there is no storage space available, contact IBM support.
Analyzing GPFS logs
Use this procedure when reviewing GPFS log entries.
Note: Contacting IBM support is recommended for any analysis of GPFS log
entries.
1. Log into the appropriate file module using root privileges.
2. Review the log file /var/adm/ras/mmfs.log.latest. The log file is ordered from
oldest to newest, so the end of the log has the latest GPFS information.
Note: The GPFS log is a complex raw log file for GPFS. If you are unable to
understand the conditions listed in the log, contact IBM support.
Synchronizing time on the file modules
Use this information to synchronize the time on all Storwize V7000 Unified file
module.
Synchronizing the time on all the file module can help as you start troubleshooting
because the timestamps on the logs then indicate whether you have concurrent,
legitimate results.
You can ensure that the Storwize V7000 Unified, Active Directory (AD), Kerberos,
and other servers are synchronized with a valid Network Time Protocol (NTP)
source. This is important both for log checking and because if the cluster falls
behind the correct time, Kerberos tickets, for example, can expire and then no one
can access the cluster. For the Storwize V7000 Unified file module, the ntpq –p
command shows you which server is used for synchronization and any peers and
a set of data about their status. The * in the first column indicates that the local
clock is used for synchronization.
# ntpq -p
remote refid st t when poll reach delay offset jitter
==============================================================================
*machine.domain.i 9.19.0.220 2 u 269 1024 377 0.659 -0.115 0.164
+machine.domain.i 9.19.0.220 2 u 992 1024 377 1.380 0.337 0.564
LOCAL(0) .LOCL. 10 l 50 64 377 0.000 0.000 0.001
Chapter 4. File module 171
To Next Page
Loading...
Need help?
General