Skip to contentSkip to conversation

"no hosts available"

DISCUSSION Mike Marion 2 participants 2 messages Jul 25, 2008 – Jul 26, 2008 Kernel automounter

lore
Share & tools
lore
#0 lore raw

Google didn't help much, nothing I see in RH's bugzilla, and I don't understand the code enough to figure out... what are a couple ideas of what could trigger the following error on an attempted mount:

automount[32110]: mount_mount: mount(nfs): no hosts available

Note, this is across a WAN connection to a filer at another site.. so could network issues do this? Just trying to track down possible causes of periodic failures to mount.

Will attempt to catch on a host with debug log vs verbose.

Signature · 7 lines
Mike Marion-Unix SysAdmin/Staff IT Engineer-http://www.qualcomm.com
[Joe just pulled Antonio away from a beautiful woman he was talking to, to move
a cabinet a tiny bit]
Antonio: "We barely moved it!"   Joe: "Yeah I know, but it was askew!  Isn't it 
amazing how something like that can throw off your whole day?"
Antonio: "Yes, and you know I only wish I could stay and slap you senseless.. 
but there's a goddess waiting for me in the terminal." -- Wings
maintainer
#1 lore raw
Quoted 9 lines · Mike Marion

Google didn't help much, nothing I see in RH's bugzilla, and I don't understand the code enough to figure out... what are a couple ideas of what could trigger the following error on an attempted mount:

automount[32110]: mount_mount: mount(nfs): no hosts available

Note, this is across a WAN connection to a filer at another site.. so could network issues do this? Just trying to track down possible causes of periodic failures to mount.

Ahhh .. you've spotted a bug.

But it also depends on what patches are included in your autofs build.

Looking at the code I see there is a bug when setting the RPC ping timeout for remote hosts. But there is a patch that eliminates the ping check for map entries don't have multiple hosts. So it may be something else.

Anyway, this patch should apply, give it a try.

autofs-5.0.3 - fix proximity other rpc ping timeout

From: Ian Kent <[email protected]>

The timeout for the RCP ping for hosts that are on a network other than the local subnet or network is mistakenly set quite short. This can lead to unexplained intermittent failures for hosts on remote networks.
---

 modules/replicated.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)
modules/replicated.c +2 -2
index efbe6b4..925f641 100644
@@ -552,7 +552,7 @@ static int get_vers_and_cost(unsigned logopt, struct host *host,
 
 	if (host->proximity == PROXIMITY_NET)
 		timeout = RPC_TIMEOUT * 2;
-	else if (host->proximity == PROXIMITY_NET)
+	else if (host->proximity == PROXIMITY_OTHER)
 		timeout = RPC_TIMEOUT * 8;
 
 	rpc_info.host = host->name;
@@ -609,7 +609,7 @@ static int get_supported_ver_and_cost(unsigned logopt, struct host *host,
 
 	if (host->proximity == PROXIMITY_NET)
 		timeout = RPC_TIMEOUT * 2;
-	else if (host->proximity == PROXIMITY_NET)
+	else if (host->proximity == PROXIMITY_OTHER)
 		timeout = RPC_TIMEOUT * 8;
 
 	rpc_info.host = host->name;