New NetApp: Difference between revisions

From CSCWiki
Jump to navigation Jump to search
(Add documentation about increasing inodes)
Line 167: Line 167:


Current schedules can be viewed by running <code>snap sched &lt;volume&gt;</code>.
Current schedules can be viewed by running <code>snap sched &lt;volume&gt;</code>.

=== inodes ===

The number of inodes can be increased with the command:

<pre>maxfiles $VOLUME $NEW_VALUE</pre>

It is not possible to decrease the number of inodes.

Revision as of 15:19, 28 April 2018

At some point in 2017, CSCF and MFCF donated us their FASXXXX NetApp filers. These filers are to replace the FAS3000 filers currently in use.

Additionally, since we were approaching maximum disk capactiy, the Math Endowment Fund funded a new 24x2TB disk shelf to go with the new filers.

NetApp Support + Documentation

As the filers were decommishioned by both CSCF and MFCF, there is no support of the filers.

Official NetApp documentation is available at https://csclub.uwaterloo.ca/~syscom/netapp-docs/.

At one point, we had access to full information about the NetApp filers on the NetApp support site. At some point, unfortunately, that stopped working. The information provided includes the license keys. We have a copy of the license keys for one of the filers (FS00) but not the other. Someone should ask CSCF or MFCF if they have this information recorded somewhere.

Physical Installation

Both of the NetApp filers are installed in the MC 3015 machine room. One filer and two disk shelves are located in rack E. The other filer was installed in rack F.

For simplicity, we decided to only use of the of the filers. We haven’t decided yet what to do with the other filer.

Networking

FS00 is connected via two 1gbps to mc-rt-3015-mso-a using LACP. Therefore, traffic should be balance between the two connections. If one of the connections goes down, the NetApp will continue to function with just the one connection.

Power

It is important that we keep the NetApp filer + disk shelves running as long as possible. At the time of installation, the UPS in rack E (mc-3015-e1-ups1) was dedicated for critical services (networking, network file shares and web hosting).

Configuration

You can SSH into the NetApp from dextrose by running ssh -oKexAlgorithms=+diffie-hellman-group1-sha1 -oCiphers=+3des-cbc root@fs00.csclub.uwaterloo.ca.

If you need information about the NetApp, run sysconfig -a on the NetApp.

Modifying /etc on the NetApp

The easiest way to change configuration on the NetApp is to mount its system directory on a different machine (only aspartame or dextrose are allowed to mount it).

mkdir /mnt/fs00
mount -t nfs -o vers=3,sec=sys fs00.csclub.uwaterloo.ca:/vol/vol0 /mnt/fs00

The NetApp system directory is currently mounted on dextrose, at /mnt/fs00.

Networking

The NetApp is configured in VLAN 530 (CSC Storage).

Here is the networking configuration in etc/rc:

# create lacp link
ifgrp create lacp csc_storage -b ip e0a e0b
ifconfig csc_storage inet 172.19.168.35 netmask 255.255.255.224 mtusize 1500
ifconfig csc_storage inet6 fd74:6b6a:8eca:4903:c5c::35 prefixlen 64
route add default 172.19.168.33 1
route add inet6 default fd74:6b6a:8eca:4903::1 1
routed on
options dns.domainname csclub.uwaterloo.ca
options dns.enable on
options nis.enable off
savecore

The CSC DNS servers are configured in etc/hosts:

nameserver 2620:101:f000:4901:c5c::4
nameserver 2620:101:f000:7300:c5c::20
nameserver 129.97.134.4
nameserver 129.97.18.20
nameserver 129.97.2.1
nameserver 129.97.2.2

TODO: The NetApp has a dedicated management port. We should take advantage of this and connect that directly to a machine which only the Systems Committee can access. Configuring this port should disable SSH via the non-management ports (this may need additional configuration).

Disks

There are two disk shelves connected to the FS00 NetApp.

  1. 14x136GB 10 000RPM FibreChannel disks
    • This was unused from our old NetApp system and was originally used for testing.
    • (ztseguin) I can’t remember, but I don’t think all disks are present.
  2. DS4243: 24x2TB 7 200RPM SATA disks
    • Funded by the Math Endowment Fund (MEF)
    • Purchased from Enterasource in Winter 2018

Aggregates

All aggregates are configured with RAID-DP.

Note: any other aggregate on the NetApp is for testing only.

aggr0

NetApp system aggregate. Disks assigned to this aggregate are located on the old disk shelf.

Volumes:

  • vol0

aggr_users

Aggregate dedicated to user home directories.

Volumes:

  • users

aggr_misc

Aggregate for miscellaneous purposes.

Volumes:

  • music

Volumes

Note: any other volume on the NetApp is for testing only.

vol0

NetApp system volume.

users

For user home directories. Each user is given a quota of 12GB.

Snapshots:

  • 12 hourly, 4 nightly and 2 weekly

music

For music.

Snapshots:

  • 2 nightly and 16 weekly

Exporting Volumes

In general, sec=sys should only be exported to MC VLAN 530 (172.19.168.32/27, fd74:6b6a:8eca:4903::/64). This VLAN is only connected to trusted machines (NetApp, CSC servers in the MC 3015 or DC 3558 machine rooms).

All other machines should be given sec=krb5p permissions only.

The NetApp exports are stored in /etc/exports. If you update the exports, they can be reloaded by running exportfs -r on the NetApp.

Quotas

Quotas are configured on the NetApp, in /etc/quotas.

After updating the quotas, the NetApp must be instructed to reload them:

# this will work for most quota changes
quota resize <volume>

# however, some changes might need a full re-initialization of quotas
#   note: while re-initializing, quotas will not be enforced.
quota off <volume>
quota on <volume>

Quota Reports

Users can view their current usage + quota by running quota -s on any machine.

The Systems Committee can run a report of everyone’s usage by running quota report on the NetApp.

Snapshots

Most volumes have snapshots enabled. Snapshots only use space when files change contained within them change (as it’s copy on write).

Snapshots are available in a special directory called .snapshot. This directory is available everywhere and will not show up in a directory listing (except at the volume root).

Current schedules can be viewed by running snap sched <volume>.

inodes

The number of inodes can be increased with the command:

maxfiles $VOLUME $NEW_VALUE

It is not possible to decrease the number of inodes.