General question Fedora 5?

31 posts / 0 new
Last post
#1 Fri, 04/14/2006 - 16:39
DanLong

General question Fedora 5?

Since I don't see any hope of Slack OS implementation I've decided to go with a bulky monster, Fedora 5. On downloading from the mirror site do I need all 5 ISO disks? That seems like a God Aweful lot, Slack only needs two, the others are source and such.

The mirrors only say ISOs 1-5, so which do I need? Since I have to set them to start downloading tonight.

THanks, Dan

If you can email me at support@solvnet.net

Fri, 04/14/2006 - 17:01
Joe
Joe's picture

Hey Dan,

Slack will be supported pretty soon, by the minimal installer mode (i.e. it doesn't install anything other than the Virtualmin modules and Webmin...but it does provide an optional script that does the post-configuration once you have all of the dependencies in place). A full installation will also come before the end of EA--but I'm not sure how soon that will be.

I don't want to be the cause of you're using an OS that you consider a bulky monster. ;-)

If size of the install disks is a concern, you might consider CentOS. They offer a single CD Server install iso. Works great and provides everything you need for the base to install Virtualmin on top of. And, of course, it has yum so you can install any extras you might want to install that normally comes from the other CDs. Many folks also dislike the general bleeding-edge feel of Fedora, which CentOS resolves. Packages aren't quite as new, but they are extremely stable and well-tested. I use Fedora on most of my servers and I'm generally happy with it, but for the systems that I really take seriously (Virtualmin.com), I use RHEL or CentOS (I actually prefer CentOS because I like yum better than up2date...but my host only offers RHEL, so that's what I'm using...price is not a determining factor for the mission critical systems).

--

Check out the forum guidelines!

Fri, 04/14/2006 - 19:20
DanLong

Thanks for the shout back Joe,

I've used Slack for years because it's very stable and been friendly
for our own servers and such, but not friendly for virtual
implemantation, ie can't write mail to users home, not able to run
clients web scripts if they are even slightly sloppy. It's not the room I
need per say as far as fedora, it's a matter of 3+ Gig of what the
heck are they putting on there? Kind of a window ish feeling if you
get my drift.

I already picked up the Fedora liscense but if you could change it to
centos I could go for a standard install and give it a try.

Fri, 04/14/2006 - 21:59
Joe
Joe's picture

Hey Dan,

No need to change the license. The installer is platform neutral across all supported systems. We just gather the data so we know where it's being used...and as a last reminder to buyers (difficult to miss, unlike the FAQs and other documentation) what platforms are supported.

--

Check out the forum guidelines!

Sun, 04/16/2006 - 23:33
DanLong

Ok,
I downloaded the 4 CentOS 4.3 ISOs, burned them, and tried to install. Nothing, the cd fails. Put Slackware in and it works fine. Is there a pre-install routine for the ISOs?

I'm currently downloading the bit torrent for disks 1-4 and maybe that might work

Mon, 04/17/2006 - 17:11
DanLong

Hmmmmm,
Seem's Nero burning software buried the image burner in a none obvious place, I was trying to work off of file copies. Now that CentOS is running I'm not overly impressed yet. Windowsish enough that like windows, I'm doing alot of guessing at how to do something. Documentation isn't really clear. Not really happy with the control-alt-delete to logoff.

Are there NIC issues I should know about? It recognizes my intelPro card in the network in the network config window, takes my static ISP info and even activates. But when I try to use the browser, nothing. When I check the network tools-devices I see the NIC eth0 with hardware address, active, but connection speed unkown and 0s fill the rest. How can I turn off the IPv6 support? Pinging the machine from another box shows a connection.

Sorry about the OS questions, but I'm trying to quickstart this ans see if I want to go for full liscense. Webmin has me so spoiled.

Tue, 04/18/2006 - 17:28 (Reply to #6)
Joe
Joe's picture

Hey Dan,

Just a few comments:

<i>Not really happy with the control-alt-delete to logoff.</i>

I've never heard of this. I know Ctrl-Alt-Backspace is the shutdown X key combo (which works on every X version I've ever used)...But, as far as I know, Ctrl-Alt-Del can be configured to either perform a clean reboot or nothing at all. By default, I think it only reboots when the login screen is up. Not sure what would be offensive about any particular default behavior, though, as it is configurable.

<i>Are there NIC issues I should know about? It recognizes my intelPro card in the network in the network config window, takes my static ISP info and even activates. But when I try to use the browser, nothing. When I check the network tools-devices I see the NIC eth0 with hardware address, active, but connection speed unkown and 0s fill the rest. How can I turn off the IPv6 support? Pinging the machine from another box shows a connection.</i>

Sounds like you either don't have DNS servers setup, or the route isn't correct. Pinging local machines by IP would work, but contacting the world would not.

You can check DNS lookups using the host command:

<i>host virtualmin.com</i>

And you can have a look at the route with:

<i>route -n</i>

You can also use traceroute to see where things are ending up:

<i>traceroute virtualmin.com</i>

No need to turn off IPv6. It won't do anything if you don't configure it. Just ignore it, like you would any other feature of the network stack you don't need.

--

Check out the forum guidelines!

Tue, 04/18/2006 - 15:28
DanLong

I'm really frustrated after recalling all the Centos/webmin problems. I've tried now for 3 days to get centos4 to work on my server box. I've tried numerous levels of install from minimal to full and all with the same problem. Inability to develop an internet connection. I've tried the suggestions I saw in the centos forum (no PnP, different NICs) and the same problem. The NIC is connected, active, pingable from the outside, set with a qualified hostname, ifconfig is fine, it just won't resove.

The install script goes fine up until it tries to connect to virtualmin.com then times out and dies.

Any suggestions? Alternate stable OS. It seems like RH and it's babies are so intent on being cutting edge they've left legacy behind well short of it's lifespan.

Dan

Tue, 04/18/2006 - 18:03 (Reply to #8)
Joe
Joe's picture

Hey Dan,

Further comments on your troubles:

<i>I'm really frustrated after recalling all the Centos/webmin problems. I've tried now for 3 days to get centos4 to work on my server box. I've tried numerous levels of install from minimal to full and all with the same problem. Inability to develop an internet connection. I've tried the suggestions I saw in the centos forum (no PnP, different NICs) and the same problem. The NIC is connected, active, pingable from the outside, set with a qualified hostname, ifconfig is fine, it just won't resove.</i>

First step: Stop reinstalling. It is unnecessary and a waste of your time. Any installation will provide basic networking capability. After that you can install any additional packages or other stuff you need using yum, very easily.

I've never seen any sort of problem with Intel NICs and CentOS. I fear you're making this step way more complicated than it needs to be.

Run <i>system-config-network</i> and follow the prompts. Save it. This will give you a basic network configuration, and if the values you gave it are correct, it <i>will work</i>. Restart the network service:

<i>service network restart</i>

Now, turn off the firewall temporarily for testing:

<i>service iptables stop</i>

Progressively more complicated tests of connectivity and configuration:

<i>ping some.local.ip</i>

<i>ping local.router.ip</i>

<i>ping 70.86.4.238</i>

<i>host virtualmin.com</i>

<i>ping virtualmin.com</i>

Assuming all of these work, you have working network connectivity. If any of them fail, let me know which one and I'll tell you how to fix it. Any one of them is extremely easy to fix. Just gotta isolate the problem.

<i>The install script goes fine up until it tries to connect to virtualmin.com then times out and dies.</i>

Yes, if the network isn't properly configured, there's no way the installer is going to work.

<i>Any suggestions? Alternate stable OS. It seems like RH and it's babies are so intent on being cutting edge they've left legacy behind well short of it's lifespan.</i>

That hasn't been my experience at all with RHEL/CentOS. It's definitely not bleeding edge (though Fedora certainly is), and exceedingly stable. Network configuration problems are just that...configuration problems. Easily remedied, assuming the hardware is well-supported. Intel NICs are very well supported. Once we isolate which part of your network settings aren't setup right, it'll be easy to fix.

Oh, yeah, once you're finished testing without the firewall, turn it back on with <i>service iptables start</i> and then run all the tests again. Should be fine--the default firewall isn't too strict to be useful, though we don't yet automatically open the ports we need in the installer--but if not, let me know and I'll post the rules you'll need to fix whatever problems pop up.

--

Check out the forum guidelines!

Tue, 04/18/2006 - 17:56
DanLong

Howdy,
Control-alt-delete was because I couldn't find my way out. Apparently the first couple installs put in Gnome which I'm unfamiliar with. I did get a KDE install then I figured that out, but the notion of working through x-window for your session was getting to me. THe last attempt, with the absolutely no frills server install left me in a shell, then I knew what to do.

I read alot of centos forum postings with the same problem and the solutions ranged from turning off PnP (never a prob before) and replacing the NIC till you got one that works. I know both of the NICs I tried work as they come from running servers.

For whatever unknown reason CentOS won't talk packets through the NIC. The IP is good, hostname, both of the nameservers are running, 4 other slack boxes, I even throw a nameserver from Bell South into the list to make sure I have one out of the loop.

There just is no routing. For a server install it doesn't remotely act like one. Apache and Bind don't start up, I had to install webmin from a disc burn just so that I could do the basic functions to get it running ( no learning curve required). The box is only 4 years old, so not old for a server, with a Tyan board. I haven't had issues with Slack or FreeBSD so I'm in a dither.

I'd go with FreeBSD but they're not good on threading and disk quotas. It can't be this bad, anything I might be missing, besides the elementary of course.

Tue, 04/18/2006 - 18:02 (Reply to #10)
Joe
Joe's picture

Hey Dan,

Let's run through the test steps I posted, and see what we get. It does sound like a routing configuration issue, which is probably easy to fix.

--

Check out the forum guidelines!

Tue, 04/18/2006 - 18:04 (Reply to #11)
Joe
Joe's picture

Just realized I didn't actually post the steps...They were sitting unposted in another tab. Posted now. ;-)

--

Check out the forum guidelines!

Tue, 04/18/2006 - 18:52
DanLong

Thanks Joe,

Here's what I have this with the simple install (no kde)

service network restart
shutting down interface eth0
shutting down loopback interface
setting network parameters:
bringing up loopback interface: iptables C ..some stuff
bringing up interface eth0: iptables C...some stuff

service iptables stop

ping solvnet.net
ping: unknown host solvnet.net

ping 70.86.4.238
from 69.128.106.188 icmp_seq=1 destination host unreachable
etc
etc

That was with the installed config

Run system-config-network and it gets weird.
name eth0
device eth0
DHCP --can't change
static IP 69.128.106.189
netmask 255.255.255.248
default gateway 69.128.106.185

save it the run service network restart

shutting down loopback interface
setting network parameters:
bringing up loopback interface: iptables C ..some stuff

that's it, no eth0
Now before the change it would hang when bringing up the interface

That's it, beyond me for sure because the network tools kept coming
up showing the card name and all.

Thanks,
Dan

Tue, 04/18/2006 - 21:41 (Reply to #13)
Joe
Joe's picture

Hey Dan,

Sounds like the interface is configured not to start on boot. A network restart only brings up interfaces configured to start on boot.

You can bring it up manually with:

<i>ifup eth0</i>

But the real solution is to set it to come up on boot in the network configuration. It's one of the options you can select with the network configuration tool.

If the ifup eth0 command gives any errors or anything, I'll need to see them.

--

Check out the forum guidelines!

Wed, 04/19/2006 - 00:05
DanLong

yikes,

ifup eth0 returns no device eth0

I reinstalled the server normally and get the same results, with the exception that now dev eth0 exists. ifconfig shows the correct Ip and routing, kde shows the correct hostname with the session starts.

Went to SU term typed system-config-network and got the kde graphical network control panel, and the realtek NIC shows up by name.

I changed the cable, went through the symantics again same results. Did whereis named to start bind and still nothing resolves. From the outside I should get the apache page but not.

Wed, 04/19/2006 - 01:48 (Reply to #15)
Joe
Joe's picture

Hey Dan,

So it's not an Intel NIC, but a Realtek, which is probably OK, as I believe Realtek NICs are also pretty well supported. I'm not as trusting of Realtek as Intel, but I don't think I've ever had any serious issue with them.

The &quot;no device eth0&quot; probably tells me one of two things:

You've switched out the card with a different type since the last time Kudzu configured the network cards. Set kudzu to run on boot (this is ordinarily the default, I think, but it might be turned off in some circumstances) and reboot. It should ask you if you want to configure the new hardware.

OR

The NIC is not the only one in the system (maybe an onboard NIC or a wireless NIC or something is also present).

I suppose the other option is your BIOS is extremely wonky. ;-)

So, we just have to step a bit further into the system to fix the problem.

Find out what NIC you have plugged in by using mii-tool (you might have to install this from your CD). Just run it from a shell prompt, as root:

mii-tool

It'll look something like this:

[[joe@delilah x86_64]]$ sudo mii-tool
eth0: negotiated 100baseTx-FD, link ok

If you have other (unplugged) NICs, they'll also show up.

I'm really surprised you're running into weirdness like this, if you've had Linux running on this system in the past without similar issues. I've rarely seen differences to something so basic as network card ordering and such between different Linux versions (and I have a lot of them running here now for Virtualmin development).

--

Check out the forum guidelines!

Wed, 04/19/2006 - 11:10
DanLong

Hi Joe,
After all that yesterday I got frustrated and tried installing FreeBSD 6 and that was a debacle, It destroyed the HDs sizing them all down to 138Gig max. But then I speak Slack now, not BSD So that was a fast died venture.

Tried a single ISO disc server install of CentOS 4.2 to no avail so I'm back to a standard 4.3 server install plus Apache, Bind and MySQL. Reinstalled the Intel EthernetPro 100 NIC, which shows up fine and active in the network tools panel. Everything we tried yesterday is duplicated again.

I've installed webmin 1.270 so I can get around in a familiar environment.

I did Mii-tool in the command line and got

eth0: no autonegotiation, 10baseT-hd, link ok

more processes than I'm used to but it runs KDE ALWAYS

anything I should look for in the extended internet services or processes? I feel like somehow localhost is my primary transport.

I'm sure that it's something really simple that RH users assume but other flavors don't consider.

Thanks,
Dan

Wed, 04/19/2006 - 13:19
DanLong

Hmmmmm,
As I'm checking, this is a running slack server setting in the webmin routing and gateway module

Active Routes
Destinaton Gateway Netmask Interface
69.128.106.184 None 255.255.255.248 eth0
127.0.0.0 None 255.0.0.0 lo
Default Route 69.128.106.185 eth0

the CentOS gives:

Active Routes
Destinaton Gateway Netmask Interface
69.128.106.184 None 255.255.255.248 eth0
169.254.0.0 None 255.255.0.0 eth0
Default Route 69.128.106.185 eth0

I don't know where that middle route comes from, but notice there is no localhost route.

I also forgot to mention that on boot the system hangs up when it gets to smartd whatever that is.

Wed, 04/19/2006 - 20:40 (Reply to #18)
Joe
Joe's picture

Hey Dan,

The routes look fine. I'm not sure why this wouldn't work. (And the localhost route isn't needed. The 169.254.0.0 is normal.)

So you said you can ping locally? Pinging your router works?

Actually, now that you've finally gotten eth0 up (as I must assume wasn't working before due to the errors you posted yesterday), why not run through those test steps again. We didn't get any useful information when last you ran them, because eth0 wasn't up.

Also useful would be the output of ifconfig.

--

Check out the forum guidelines!

Wed, 04/19/2006 - 20:59
Joe
Joe's picture

Oh, yeah, smartd is the S.M.A.R.T. monitoring daemon. If your system disks or controller doesn't support it, it'll fail (or if it's not /dev/hda, which is the default). You can either configure it correctly in /etc/smartd.conf or you can disable it with:

chkconfig --level 345 smartd off

It's normally a harmless error, but I have seen it hang the system on some buggy Adaptec (and maybe other) controllers.

--

Check out the forum guidelines!

Wed, 04/19/2006 - 21:35
DanLong

Well,
We having a working model, all that is except the external routing. I can ping the box but can't ping out. Hopefully ifconfig will give you an idea, after further review it's not &quot;just fine&quot;

[[root@hosta webmin-1.270]]# ifconfig
eth0 Link encap:Ethernet HWaddr 00:02:B3:0A:85:97
inet addr:69.128.106.189 Bcast:69.128.106.191 Mask:255.255.255.248
inet6 addr: fe80::202:b3ff:fe0a:8597/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:22 txqueuelen:1000
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:20671 errors:0 dropped:0 overruns:0 frame:0
TX packets:20671 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:18501706 (17.6 MiB) TX bytes:18501706 (17.6 MiB)

Thu, 04/20/2006 - 16:06 (Reply to #21)
Joe
Joe's picture

Hey Dan,

So eth0 is definitely up. Strangely I'm only seeing collisions on this interface...and that aint right.

Could there be another device on your network with the same IP or MAC address?

What appears in the kernel log when you try to ping out? (This can be seen by typing &quot;dmesg&quot; at the command line.)

--

Check out the forum guidelines!

Thu, 04/20/2006 - 17:48
DanLong

Good thing God gave me lots of hair!!

Here's this:

/etc/sysconfig/network/ifconfig-eth0:
GATEWAY=69.128.106.185
BOOTPROTO=none
TYPE=Ethernet
HWADDR=00:02:B3:0A:85:97
DEVICE=eth0
MTU=&quot;&quot;
NETMASK=255.255.255.248
BROADCAST=69.128.106.191
IPADDR=69.128.106.189
NETWORK=69.128.106.184
ONBOOT=yes

/resolv.conf:
search solvdns.net
nameserver 216.126.128.40
nameserver 4.2.2.1
nameserver 216.165.129.157

/network-scripts/ifconfig-eth0
GATEWAY=69.128.106.185
BOOTPROTO=none
TYPE=Ethernet
HWADDR=00:02:B3:0A:85:97
DEVICE=eth0
MTU=&quot;&quot;
NETMASK=255.255.255.248
BROADCAST=69.128.106.191
IPADDR=69.128.106.189
NETWORK=69.128.106.184
ONBOOT=yes

When I manually changed BOOTPROTO to static on all occasions and rebooted It froze the boot at CUPS, NFS and such. I never did get it to route.

Just to check the card I reinstalled Slack 10.2 ( twice after realizing that GRUB was living in the MBR) I went right away to startx, opened Konquerer, tyoed in webmin.com and boom, webmin shows up. So the card isn't fried.

I'll give it one last gasp, trying to make sure that CUPS and NFS are not in it (who needs printing on a server??)

Thu, 04/20/2006 - 19:14 (Reply to #23)
Joe
Joe's picture

Hey Dan,

CUPS and NFS would only freeze (and not really freeze...they would timeout and eventually fail and the boot would continue after some period of time) if you don't have any interfaces (i.e. not even loopback) or if they have been configured explicitly to use other non-loopback interfaces. I'm not sure how you'd end up in that situation. If they actually freeze the system then I'm pretty sure your hardware has some kind of pretty serious problem.

I think I'd still like to see the kernel log output. It seems like something serious is wrong on your box, and I don't know that configuration is going to change that. dmesg might give you some clues about where things are going awry.

--

Check out the forum guidelines!

Thu, 04/20/2006 - 18:56
DanLong

It was no go with the 4.2 server iso. Same problem. Thought I would try letting it install as a DHCP device but all that did was hang up the boot process.

I'm DLing the 3.7 server torrent and will give that a try. Otherwise it's a bug and I'll send in the motherboard and NIC info.

tapping fingers waiting for the Slack version............

Thu, 04/20/2006 - 18:58
DanLong

OH,
How does one mount and access the CDROM or floppy in barebones centOS?

Thu, 04/20/2006 - 20:15
DanLong

LOL, freeze is the word for the impatient. The timeouts were incredibly long, in the area of 15-20 minutes if I could wait that long.

Thu, 04/20/2006 - 20:15
DanLong

I did the mesg and got this eth related stuff

e100:Intel pro/100 driver 3.4.8-k2-napi
e100:eth0:e100_watchdog: linkup, 10 mps, halfduplex
eth0: no IPv6 router present

net: registering protcol family 2
IP: routing cache hash table of 4096 buckets, 64 Kbytes
TCP: hash tables configured (established 131072 bind 43690)
Initialized IPsec netlink socket
net: registering protcol family 1
net: registering protcol family 17

net: registering protcol family 10
Disabled privacy extensions on device xxxxxxxx (lo)
IPv6 over IPv4 tunneling driver

and then these divert statements:

divert: allocating diver_blk for eth0

divert: not allocating divert_blk for nonethernet device sit0

What is sit0??

The rest was standard fare boot items and probes

half duplex I've never seen but shouldn't bother, I'm familiar with this IPv6 over IPv4 tunneling though.

Thu, 04/20/2006 - 22:45
DanLong

well, CentOS 3.7 server installed and pings outside the box. I ran your virtmin.sh. What is the username and password for webmin?

Thu, 04/20/2006 - 22:47
DanLong

skip the last post ;-)

Fri, 04/21/2006 - 08:54
DanLong

I have virtminPro up and running now and am fairly impressed, having been using Virtualmin for a few years now the extras deserve kudos.

As I'm tweaking I have to ask, do I want to try to upgrade to CentOS4? The most obvious thing I noticed on install of 3 was the user option to choose to run all traffic through eth0 ( had to actually select the device, though it was the only choice)

Still have no idea why CentOS 4 refused to route.

Not stop is throwing AWBS into the works. Anything I should know in advance?

Topic locked