FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Redhat > Fedora User

 
 
LinkBack Thread Tools
 
Old 06-20-2008, 07:45 PM
"Matt Nicholson"
 
Default problems w/ net (http) install

Greetings everyone,

So, I'm trying to setup a local server for some net-installs I hope to do with a kickstart file. I am, however, running into an issue.

I have a copy of the fedora 9 install media on the web server that the install will be pulled from, and everything is in tip top shape. This server is actually a fairly new Xserve, and I am using it simply because it is available, has to disk-space and bandwidth, and is a pretty fast system for multiple systems to kickstart aganst. I would rather be doing this off a Fedora/RHEL server, but, this is what I have for the time being.


Anyways, I've rsync'd the install media to the server, and its accessible, however, durring the install, I always get a file or two (sometimes different, sometimes the same), that anaconda spits back at me, saying it could not find/read the file, make sure its not currupted, etc etc etc. I can reboot, or retry, and retry always works, that is, until it hit the next file ti doesn't like. I get about 3-4 of these per install, EVERY TIME. I've checked, the files are there, they are the right size, I've even done an MD% of them and they match their sources. I even re-rsync'd the whole thing a few times.If this is a one time deal, I wouldn't mind, but I need to be able to basically start an install (via kickstart) and walk away.


Now, normally, I would just say forget it, and do it over FTP, but FTP on this Xserver is very, very slow, and my installs, while succeeding without error, are about 10 times longer with the same package set. Also the network alyout means NFS is off the table as well.


Any ideas? I would love any insight.

Thanks!

Matt

--
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
 
Old 06-20-2008, 08:16 PM
Rick Stevens
 
Default problems w/ net (http) install

Matt Nicholson wrote:

Greetings everyone,

So, I'm trying to setup a local server for some net-installs I hope to do
with a kickstart file. I am, however, running into an issue.

I have a copy of the fedora 9 install media on the web server that the
install will be pulled from, and everything is in tip top shape. This server
is actually a fairly new Xserve, and I am using it simply because it is
available, has to disk-space and bandwidth, and is a pretty fast system for
multiple systems to kickstart aganst. I would rather be doing this off a
Fedora/RHEL server, but, this is what I have for the time being.

Anyways, I've rsync'd the install media to the server, and its accessible,
however, durring the install, I always get a file or two (sometimes
different, sometimes the same), that anaconda spits back at me, saying it
could not find/read the file, make sure its not currupted, etc etc etc. I
can reboot, or retry, and retry always works, that is, until it hit the next
file ti doesn't like. I get about 3-4 of these per install, EVERY TIME. I've
checked, the files are there, they are the right size, I've even done an MD%
of them and they match their sources. I even re-rsync'd the whole thing a
few times.If this is a one time deal, I wouldn't mind, but I need to be able
to basically start an install (via kickstart) and walk away.

Now, normally, I would just say forget it, and do it over FTP, but FTP on
this Xserver is very, very slow, and my installs, while succeeding without
error, are about 10 times longer with the same package set. Also the network
alyout means NFS is off the table as well.

Any ideas? I would love any insight.


Are the the failling packages big ones? If so, you may want to turn up
the wick on the TCP keepalives, client timeouts and make sure you're
not limiting the size of a "served" page on the web server. It's
possible that the connection is being terminated in the middle of the
file transfer. A retry would pick up the parts of the file that didn't
get there the first time.

Just an idea.
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer rps2@nerd.com -
- Hosting Consulting, Inc. -
- -
- When all else fails, try reading the instructions. -
----------------------------------------------------------------------

--
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
 
Old 06-20-2008, 08:36 PM
Craig White
 
Default problems w/ net (http) install

I. On Fri, 2008-06-20 at 15:45 -0400, Matt Nicholson wrote:
> Greetings everyone,
>
> So, I'm trying to setup a local server for some net-installs I hope to
> do with a kickstart file. I am, however, running into an issue.
>
> I have a copy of the fedora 9 install media on the web server that the
> install will be pulled from, and everything is in tip top shape. This
> server is actually a fairly new Xserve, and I am using it simply
> because it is available, has to disk-space and bandwidth, and is a
> pretty fast system for multiple systems to kickstart aganst. I would
> rather be doing this off a Fedora/RHEL server, but, this is what I
> have for the time being.
>
> Anyways, I've rsync'd the install media to the server, and its
> accessible, however, durring the install, I always get a file or two
> (sometimes different, sometimes the same), that anaconda spits back at
> me, saying it could not find/read the file, make sure its not
> currupted, etc etc etc. I can reboot, or retry, and retry always
> works, that is, until it hit the next file ti doesn't like. I get
> about 3-4 of these per install, EVERY TIME. I've checked, the files
> are there, they are the right size, I've even done an MD% of them and
> they match their sources. I even re-rsync'd the whole thing a few
> times.If this is a one time deal, I wouldn't mind, but I need to be
> able to basically start an install (via kickstart) and walk away.
>
> Now, normally, I would just say forget it, and do it over FTP, but FTP
> on this Xserver is very, very slow, and my installs, while succeeding
> without error, are about 10 times longer with the same package set.
> Also the network alyout means NFS is off the table as well.
>
> Any ideas? I would love any insight.
----
I'd be curious about why the network layout means that NFS is off the
table but HTTP is on the table.

Anyway, are you 'loop' mounting the ISO files? Is there something that
delays reading the files?

Is Energy saving allowing the hard drive to spin down on the XServer?
(Mac's sometimes default to sleep modes with hard drive spin down which
would be a mistake for a server).

What OS is on the X-Serve? Are there errors in the web server logs on
the X-Serve?

Craig

--
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
 
Old 06-20-2008, 09:36 PM
"Matt Nicholson"
 
Default problems w/ net (http) install

Craig,

For your 4 points:

NFS isn't strictly off the table, but the system hosting this install tree will need to be accessible from alot of system across a large number of subnets/VLAN's. I would rather have port 80 open to these nets/the world than NFS, but then again I can just make it an "ro" export. Something to try next week.


I'm not mounting the iso's, but rather have full fledged, rsync'd copies of the install tree, local on disk.

No energy saving on the Xserve. It doesn't powerdown/spin down at all, ever.


The Xserve is running Leopard Server, 10.5.3. Unfortunatly, no erros in the logs. Everything looks normal.

And Rick,

Nope the packages aren't big ones, fairly standard, 1MB-ish packages, although the packages do change. The keep alive is set at 300 seconds, which = 5 minutes. The thing is, this is all happening while anaconda is preparing to install (ie, not when its acctually downloading and installing the rpms, the set jsut before that starts). It zips right though until it hits one of these files. If it wasn't interupted, the whole thing could finish in maybe 1 minute, if not less, so I don't think timeouts are an issue. I've even up'd the number of conenction Apache allows, and the nubmer of servers it spawns, jsut incase anaconda was hammering it with too many requests.


Matt

On Fri, Jun 20, 2008 at 4:36 PM, Craig White <craigwhite@azapple.com> wrote:

* * I. On Fri, 2008-06-20 at 15:45 -0400, Matt Nicholson wrote:

> Greetings everyone,

>

> So, I'm trying to setup a local server for some net-installs I hope to

> do with a kickstart file. I am, however, running into an issue.

>

> I have a copy of the fedora 9 install media on the web server that the

> install will be pulled from, and everything is in tip top shape. This

> server is actually a fairly new Xserve, and I am using it simply

> because it is available, has to disk-space and bandwidth, and is a

> pretty fast system for multiple systems to kickstart aganst. I would

> rather be doing this off a Fedora/RHEL server, but, this is what I

> have for the time being.

>

> Anyways, I've rsync'd the install media to the server, and its

> accessible, however, durring the install, I always get a file or two

> (sometimes different, sometimes the same), that anaconda spits back at

> me, saying it could not find/read the file, make sure its not

> currupted, etc etc etc. I can reboot, or retry, and retry always

> works, that is, until it hit the next file ti doesn't like. I get

> about 3-4 of these per install, EVERY TIME. I've checked, the files

> are there, they are the right size, I've even done an MD% of them and

> they match their sources. I even re-rsync'd the whole thing a few

> times.If this is a one time deal, I wouldn't mind, but I need to be

> able to basically start an install (via kickstart) and walk away.

>

> Now, normally, I would just say forget it, and do it over FTP, but FTP

> on this Xserver is very, very slow, and my installs, while succeeding

> without error, are about 10 times longer with the same package set.

> Also the network alyout means NFS is off the table as well.

>

> Any ideas? I would love any insight.

----

I'd be curious about why the network layout means that NFS is off the

table but HTTP is on the table.



Anyway, are you 'loop' mounting the ISO files? Is there something that

delays reading the files?



Is Energy saving allowing the hard drive to spin down on the XServer?

(Mac's sometimes default to sleep modes with hard drive spin down which

would be a mistake for a server).



What OS is on the X-Serve? Are there errors in the web server logs on

the X-Serve?



Craig



--

fedora-list mailing list

fedora-list@redhat.com

To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list



--
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
 
Old 06-20-2008, 10:40 PM
"Matt Nicholson"
 
Default problems w/ net (http) install

Okay, so, I just downloaded my custom bootcd and kickstarted a VM in virtualbox here @ home {(everything else is @ work). And yep. no problems doing the install over http. So, now I'm not sure if it was a symtom of the VM i was doing my test installs in @ work, if the speed (1GB) of the connection to the system was over running things (I'm on cable @ home, so its fast, but not that fast), so what, but, it doesn't seem to be a "Fedora" problem now so much as problem with my setup.


I must admit, I'm a Ubuntu/Debian guy, but do a lot of RHEL @ work. I haven't had much interaction with the Fedora Community, and this was my first email to this list. Its a really good impression, what with the helpful suggestions and all, on such an odd issue. I'm sure I'll return.


Thanks,
Matt

On Fri, Jun 20, 2008 at 5:36 PM, Matt Nicholson <sjoeboo@sjoeboo.com> wrote:

Craig,

For your 4 points:

NFS isn't strictly off the table, but the system hosting this install tree will need to be accessible from alot of system across a large number of subnets/VLAN's. I would rather have port 80 open to these nets/the world than NFS, but then again I can just make it an "ro" export. Something to try next week.



I'm not mounting the iso's, but rather have full fledged, rsync'd copies of the install tree, local on disk.

No energy saving on the Xserve. It doesn't powerdown/spin down at all, ever.



The Xserve is running Leopard Server, 10.5.3. Unfortunatly, no erros in the logs. Everything looks normal.

And Rick,

Nope the packages aren't big ones, fairly standard, 1MB-ish packages, although the packages do change. The keep alive is set at 300 seconds, which = 5 minutes. The thing is, this is all happening while anaconda is preparing to install (ie, not when its acctually downloading and installing the rpms, the set jsut before that starts). It zips right though until it hits one of these files. If it wasn't interupted, the whole thing could finish in maybe 1 minute, if not less, so I don't think timeouts are an issue. I've even up'd the number of conenction Apache allows, and the nubmer of servers it spawns, jsut incase anaconda was hammering it with too many requests.



Matt

On Fri, Jun 20, 2008 at 4:36 PM, Craig White <craigwhite@azapple.com> wrote:


* * I. On Fri, 2008-06-20 at 15:45 -0400, Matt Nicholson wrote:

> Greetings everyone,

>

> So, I'm trying to setup a local server for some net-installs I hope to

> do with a kickstart file. I am, however, running into an issue.

>

> I have a copy of the fedora 9 install media on the web server that the

> install will be pulled from, and everything is in tip top shape. This

> server is actually a fairly new Xserve, and I am using it simply

> because it is available, has to disk-space and bandwidth, and is a

> pretty fast system for multiple systems to kickstart aganst. I would

> rather be doing this off a Fedora/RHEL server, but, this is what I

> have for the time being.

>

> Anyways, I've rsync'd the install media to the server, and its

> accessible, however, durring the install, I always get a file or two

> (sometimes different, sometimes the same), that anaconda spits back at

> me, saying it could not find/read the file, make sure its not

> currupted, etc etc etc. I can reboot, or retry, and retry always

> works, that is, until it hit the next file ti doesn't like. I get

> about 3-4 of these per install, EVERY TIME. I've checked, the files

> are there, they are the right size, I've even done an MD% of them and

> they match their sources. I even re-rsync'd the whole thing a few

> times.If this is a one time deal, I wouldn't mind, but I need to be

> able to basically start an install (via kickstart) and walk away.

>

> Now, normally, I would just say forget it, and do it over FTP, but FTP

> on this Xserver is very, very slow, and my installs, while succeeding

> without error, are about 10 times longer with the same package set.

> Also the network alyout means NFS is off the table as well.

>

> Any ideas? I would love any insight.

----

I'd be curious about why the network layout means that NFS is off the

table but HTTP is on the table.



Anyway, are you 'loop' mounting the ISO files? Is there something that

delays reading the files?



Is Energy saving allowing the hard drive to spin down on the XServer?

(Mac's sometimes default to sleep modes with hard drive spin down which

would be a mistake for a server).



What OS is on the X-Serve? Are there errors in the web server logs on

the X-Serve?



Craig



--

fedora-list mailing list

fedora-list@redhat.com

To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list





--
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
 

Thread Tools




All times are GMT. The time now is 12:40 PM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright 2007 - 2008, www.linux-archive.org