X-Git-Url: https://git.gag.com/?a=blobdiff_plain;f=debian%2Famanda-server.README.Debian;h=c51849cdb6dc7fa5747aae3c8ffa5b10a500cb68;hb=322fc055c22c414a2baf9159b8eb8e74a84f70c7;hp=809b966f89199517ea58d27b88d05b357f995c60;hpb=afaa71b3866b46b082b6c895772e15b36d8865fe;p=debian%2Famanda

diff --git a/debian/amanda-server.README.Debian b/debian/amanda-server.README.Debian
index 809b966..c51849c 100644
--- a/debian/amanda-server.README.Debian
+++ b/debian/amanda-server.README.Debian
@@ -47,12 +47,10 @@ machine.  On Debian systems, this is usually linked to /etc/amandahosts, but
 if you have clients running other operating systems you'll have to find the
 right place yourself.
 
-
 NOTE:  If you want to backup your server too, you need to treat it as both a
 client *and* a server, see /usr/share/doc/amanda-client/README.Debian for 
 more information.
 
-
 For more complex setups, consult the manpages and the extra documents and
 example files in /usr/share/doc/amanda-common, or look at www.amanda.org on
 the web.
@@ -72,7 +70,6 @@ his experience using ftape:
     pre-install zftape /sbin/swapout 5
     post-install zftape /bin/ftmt -f /dev/nzqft0 setblk 32768
 
-
 To ease use with firewalls, Amanda has been built with options restricting the
 use of TCP to the port range 50000-50100,  and UDP to the port range 840-860.
 
@@ -124,3 +121,24 @@ info on this issue see:
 
 	http://forums.zmanda.com/archive/index.php/t-591.html
 
+- - - - -
+
+If you see things like this in dmesg:
+
+    INFO: task ammt:9839 blocked for more than 120 seconds.
+    "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
+
+This information from Marc Schaefer may relevant:
+
+> AFAIK it's because mtio operations are done in D state (uninterruptible
+> sleep).
+> 
+> And except for mtio operations which may last more than two minutes,
+> nothing in the usual kernel operation will last that long.
+> 
+> Solutions:
+>	- either fix mtio to be interruptible (which may well cause problems
+>	  because e.g. SCSI reselection might happen later)
+>	- use the proposed work-around
+>	- ignore those messages in your logcheck.
+