A NetApp Technical Diary

*** WARNING *** Hammer is a system level stress test.
CPU utilization may approach 100 percent
with very high disk IO rate when hammer is run.
This may adversely affect system performance
and may cause loss of filer service.
Hammer is an undocumented, unsuppported command.

This commands just wants to be used by people like me! Another stress test tool like filersio. This doesn’t do any sort of reporting though, it just hammers the system! So use something like sysstat or stats to report on what the filer is doing.

usage: hammer [abort|pause|restart|status|
       [-f]<# Runs><fileName><# BlocksInFile> (<# Runs> == -1 runs hammer forever)|
       fill <writeSize> (use all available disk space)]

filer1*> hammer -f 5 /vol/vol0/hammer.txt 400
Mon Dec  8 12:32:02 GMT [blacksmith:warning]: blacksmith #0: Starting work.
filer1*> hammer status
blacksmith #0: file /vol/vol0/hammer.txt, 580 KB of 1600 KB writesize 4096 bytes, iteration 1 of 5 – Writing
filer1*> hammer status
blacksmith #0: file /vol/vol0/hammer.txt, 1600 KB of 1600 KB writesize 4096 bytes, iteration 1 of 5 – Writing
filer1*> hammer status
blacksmith #0: file /vol/vol0/hammer.txt, 652 KB of 1600 KB writesize 4096 bytes, iteration 2 of 5 – Writing
filer1*> hammer status
blacksmith #0: file /vol/vol0/hammer.txt, 1600 KB of 1600 KB writesize 4096 bytes, iteration 2 of 5 – Writing
filer1*> Mon Dec  8 12:32:13 GMT [blacksmith:info]: blacksmith #0: No errors detected. Stopping work

 

Undocumented Commands hammer, NetApp, ONTAP

This is quite useful for extracting information from the filer.

usage: getXXbyYY <sub-command> <name>
Where sub-command is one of
         gethostbyname_r
         gethostbyaddr_r
         netgrp
         getspwbyname_r
         getpwbyname_r
         getpwbyuid_r
         getgrbyname
         getgrbygid
         getgrlist

So we can pull out the root users information, including the encrypted password

 filer1*> getXXbyYY getpwbyname_r root
pw_name = root
pw_passwd = _J9..9aTB5SBQOW.uu96
pw_uid = 0, pw_gid = 1
pw_gecos =
pw_dir = /
pw_shell =

Supplemental:

These tools are actually very useful, and on further reading and applications I’ve realise more where they can be used. Remember the search order is defined by “/etc/nsswitch.conf”

To perform DNS lookups, similar to what you might want to achieve with nslookup, you can use this tool…

filer01> getXXbyYY gethostbyname_r b2net-dc
name: b2net-dc.b2net.lan
aliases:
addresses: 10.0.0.5 10.0.1.5 10.0.30.5

Or reverse lookups to make sure DNS works properly…

filer01*> getXXbyYY gethostbyaddr_r 10.0.0.149
name: b2net-exch03.b2net.lan
aliases:
addresses: 10.0.0.149

If you have LDAP configured and working, you can also use this tool to perform LDAP lookups.

Undocumented Commands getXXbyYY, ldap, NetApp, nslookup, ONTAP

No output and can’t seem to get anything from this. Is this the GNU debugger?

Undocumented Commands gdb, NetApp, ONTAP

 filer1*> filersio
The following commands are available; For more information
filersio help <command>
asyncio_pause           asyncio_active          spc1
status                  stop

As you expect, this is an IO generator. A nice way of testing a filer, perhaps to see what it can do, or perhaps to do some performance testing to see if there is an issue. hammer can also be used to generate IO on the filer.

 filer1*> filersio asyncio_active 50 -r 50 4 0 10m 60 5 /vol/vol0/filersio.test -create -print_stats 5
filersio: workload initiated asynchronously. Results will be displayed on the
console after completion
filersio: starting workload asyncio_active, instance 0

Read I/Os       Avg. read       Max. read       Write I/Os      Avg. write      Max write
                latency(ms)     latency(ms)                     latency(ms)     latency(ms)
6087            0               11              6216            3               981
4410            0               471             4300            5               1551
5323            0               11              5375            4               1121
5439            0               113             5379            4               1151
4354            0               105             4304            5               1674
4307            0               411             4300            5               1171
5459            0               20              5371            5               1260
5384            0               180             5379            4               1071
5439            0               211             5375            4               1011
4396            0               71              4300            5               1311

Statistics for active_active model, instance 0
Running for 60s
Total read latency(ms)  16383
Read I/Os               51687
Avg. read IOPS          861
Avg. read latency(ms)   0
Max read latency(ms)    471
Total write latency(ms) 286501
Write I/Os              51413
Avg. write IOPS         856
Avg. write latency(ms)  5
Max write latency(ms)   4891
filersio: instance 0: workload completed successfully

Undocumented Commands filersio, NetApp, ONTAP

filer1*> export_stats
ERROR: One of -p, -z, or -h required.

export_stats has three basic variants: -h, -z, and -p

-h is specified alone and produces help output such as this.

-p displays export statistics.
        -v (verbose) may be specified with -p and adds statistics descriptions
        -q (quiet) may be specified with -p and suppresses display of zero counts.

-z zeros export statistics.

-p and -z require specification of a set of modules via either:
        -a to specify all modules.
        one or more -m <modname> specifications, where:
                <modname> is one of the three-letter export module codes

The following module codes are defined:
        IAR — Information About Rules
        REQ — Request Handling
        ADI — Access Determination Interface
        AEL — Access Evaluation Layer
        ALI — Address List Interface
        AII — Address Item Interface
        ACP — Access Cache Population
        ATI — Address Type Implementation

Again, pretty much what it says on the tin! shows some detailed specific NFS stats.

“export_stats -pav” will show all stats.

Undocumented Commands export_stats, NetApp, ONTAP

Again, this is documented, but I find this tool can be useful for other purposes. It is useful for simply copying files around, regardless of permissions! It can also be used to copy files out of snapshots, even if you don’t have SnapRestore.

dd if=/vol/vol0/.snapshot/hourly.0/etc/rc of-/vol/vol0/etc/rc.old

if = input file
of = output file

Undocumented Commands dd, NetApp, ONTAP

filer*> dbg

usage: dbg to [ console | log | off ]

       dbg dump

       dbg dumpcons

       dbg level [ <module>=<level> ... ]

Debugging. This can be turned on or off from here and also you can tell it to dump to console, or dump to a file. The file is “/,dbg”.

Undocumented Commands dbg, NetApp, ONTAP

filer*> bootfs

For more information, type “bootfs help <command>”.
The following bootfs commands are available:

chkdsk         core           dir            dump           fdisk
format         info           sync           test

This command is documented clearly as it is needed to format the compact flash card if ever needed. A helpful hint though, the boot dir is usually “0a.0:1:”, so you can do “bootfs dir 0a:0:1:” to see what’s in there.

Undocumented Commands bootfs, NetApp, ONTAP

blink_on / blink_off <disk_name>

I like flashing lights!

This can actually be useful if you need to locate a disk. Don’t forget led_on and led_off, but there are alternatives.

Undocumented Commands blink_off, blink_on, NetApp, ONTAP

Pretty much exactly what it says on the tin! Gives you availability times for the system and each protocol. This can be reset with “availtime reset”.

Undocumented Commands availtime, NetApp, ONTAP