[OpenAFS] Fault tolerance in AFS
Shyh-Wei Luan
luan@almaden.ibm.com
Fri, 14 Sep 2001 11:50:54 -0700
--0__=07BBE054DFF670D68f9e8a93df938690918c07BBE054DFF670D6
Content-type: text/plain; charset=US-ASCII
Craig,
I think you meant "fs checkv" or "fs checkvolumes" here ... I think an
automatic "fs checkvol" equivalent action can be taken by the client to
find the potential new servers.
BTW, We have experimented using Fibre Channel SAN to allow the vice
partitions to be accessible from multiple AFS file servers, but with only
one of them actively serving any particular partition. When any server
fails, one of the other servers can start serving the vice partitions of
the failed server. We also tried having a standby server taking over the
failed server's network identity. Either scheme should be able to take
care of server hardware failure very well and keep the server downtime
minimal.
Shyh-Wei
Craig_Everhart@transarc.com@openafs.org on 09/14/2001 07:29:49 AM
Sent by: openafs-info-admin@openafs.org
To: "Openafs-Info (E-mail)" <openafs-info@openafs.org>
cc:
Subject: Re: [OpenAFS] Fault tolerance in AFS
AFS doesn't replicate R/W data from one server to another. You can
approximate this in various ways, either using parts of AFS (vos
dump|vos restore) or not (mirrored or dual-ported disks). Once data is
available on a server that's up, you can use vos syncvldb to point the
clients to the new data location. The clients will hear from the VLDB
within an hour or when the client does "fs checkb".
Craig
_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info
--0__=07BBE054DFF670D68f9e8a93df938690918c07BBE054DFF670D6
Content-type: text/html; charset=US-ASCII
Content-Disposition: inline
<html><body><br>
<br>
Craig,<br>
<br>
I think you meant "fs checkv" or "fs checkvolumes" here ... I think an automatic "fs checkvol" equivalent action can be taken by the client to find the potential new servers.<br>
<br>
BTW, We have experimented using Fibre Channel SAN to allow the vice partitions to be accessible from multiple AFS file servers, but with only one of them actively serving any particular partition. When any server fails, one of the other servers can start serving the vice partitions of the failed server. We also tried having a standby server taking over the failed server's network identity. Either scheme should be able to take care of server hardware failure very well and keep the server downtime minimal. <br>
<br>
Shyh-Wei<br>
<br>
<br>
<p><font size=2 color="#800080">Sent by: openafs-info-admin@openafs.org</font>
<p><font size=2 color="#800080">To: </font><font size=2>"Openafs-Info (E-mail)" <openafs-info@openafs.org></font><br>
<font size=2 color="#800080">cc: </font><br>
<font size=2 color="#800080">Subject: </font><font size=2>Re: [OpenAFS] Fault tolerance in AFS</font><br>
<br>
<br>
<br>
<tt>AFS doesn't replicate R/W data from one server to another. You can<br>
approximate this in various ways, either using parts of AFS (vos<br>
dump|vos restore) or not (mirrored or dual-ported disks). Once data is<br>
available on a server that's up, you can use vos syncvldb to point the<br>
clients to the new data location. The clients will hear from the VLDB<br>
within an hour or when the client does "fs checkb".<br>
</tt><br>
<tt> Craig<br>
_______________________________________________<br>
OpenAFS-info mailing list<br>
OpenAFS-info@openafs.org<br>
</tt><tt><a href="https://lists.openafs.org/mailman/listinfo/openafs-info">https://lists.openafs.org/mailman/listinfo/openafs-info</a></tt><br>
<br>
</body></html>
--0__=07BBE054DFF670D68f9e8a93df938690918c07BBE054DFF670D6--