e1000e startup problem
by Kevin H. Hobbs
I have a fedora 14 server here who's ethernet interface seems to take
just a bit too long to establish a link.
Whenever it starts it is inaccessible over the network, which is a real
pain for a server. I have to go over with the monitor and keyboard,
crawl under the desk...
When I try to restart the network interface with "service network
restart" I get something that looks like :
Bringing up interface eth0:
Determining IP information for eth0...[big number] ADDRCONF(NETDEV_UP):
eth0: link is not ready.....
[FAILED]
[big number] e1000e: eth0 NIC Link is Up 1000 Mbps Full Duplex...
Which sounds to me like a timeout was just a bit too short.
If I try a few times in a row suddenly it works.
[kevin@backup ~]$ uname -a
Linux backup.hooperlab 2.6.35.14-96.fc14.x86_64 #1 SMP Thu Sep 1
11:59:56 UTC 2011 x86_64 x86_64 x86_64 GNU/Linux
[kevin@backup ~]$ lspci -nn | grep -i eth
04:00.0 Ethernet controller [0200]: Intel Corporation 82573E Gigabit
Ethernet Controller (Copper) [8086:108c] (rev 03)
05:05.0 Ethernet controller [0200]: Intel Corporation 82541GI Gigabit
Ethernet Controller [8086:1076] (rev 05)
[kevin@backup ~]$ dmesg | tail
[ 1633.233297] e1000e 0000:04:00.0: irq 73 for MSI/MSI-X
[ 1633.284094] e1000e 0000:04:00.0: irq 73 for MSI/MSI-X
[ 1633.285079] ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 1642.265283] e1000e 0000:04:00.0: irq 73 for MSI/MSI-X
[ 1642.316094] e1000e 0000:04:00.0: irq 73 for MSI/MSI-X
[ 1642.317556] ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 1647.280547] e1000e: eth0 NIC Link is Up 100 Mbps Full Duplex, Flow
Control: RX/TX
[ 1647.285563] e1000e 0000:04:00.0: eth0: 10/100 speed: disabling TSO
[ 1647.291153] ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[ 1657.802008] eth0: no IPv6 routers present
12 years, 6 months
Kernel Questions (pos compiling)
by Joao Daniel Neves
Hi folks,
I have compiled kernel 3.1 rc7 under Fedora 15.
Compiling the kernel drive me into a several question.
I tought that if I remove support for hardware that I wont use it would improve performance (at least boot time ).
But booting from my new kernel (instead of the 'old' fedora kernel) take much more time.
So I discovered that my initramfs file was to large. About 112 MB. Fedora intramfs is about 15Mb.
Basically my questions are:
1) If I remove SCSI support from kernel, would it improve performace? More generally what improve performace in linux kernel. It is possible to improve performace just by removing itens?
2)Why my initramfs is too big? I unselect a lot o stuff that I wont use. It is possible to make a intramsfs of 1MB ?
12 years, 6 months
I've got an interesting problem (at least to me).....
by Kevin Martin
I've got a Toshiba laptop that I have two drives in (/dev/sda and /dev/sdb). I was having some issues with /dev/sda so I bought the
second drive, dd'd drive sda contents to drive sdb, and the removed sda and replaced it. I used sdb for the boot drive until
receiving the replacement. Upon arrival, I inserted the new drive as sda, dd'd the contents from sdb back to sda, and rebooted. It
appears that I'm still rebooting from sdb at this point (which is ok with me). What I don't understand and don't like to see is the
following:
/dev/sda11 on / type ext4 (rw,relatime,user_xattr,barrier=1,data=ordered)
/dev/sdb5 on /boot type ext4 (rw,relatime,user_xattr,barrier=1,data=ordered)
/dev/sda12 on /work type ext4 (rw,relatime,user_xattr,barrier=1,data=ordered)
/dev/sdb7 on /usr type ext4 (rw,relatime,user_xattr,barrier=1,data=ordered)
/dev/sdb8 on /var type ext4 (rw,relatime,user_xattr,barrier=1,data=ordered)
/dev/sda6 on /home type ext4 (rw,relatime,user_xattr,barrier=1,data=ordered)
/dev/sdb9 on /opt type ext4 (rw,relatime,user_xattr,barrier=1,data=ordered)
Why is Fedora (16 x64 beta in this case) cross mounting my partitions? I was halfway considering setting up a mirror for these
drives but not I'm going to have to do some very particular dd'ing to get one drive to be the "latest" files. Is this SOP or is
something odd going on here?
Thanks.
Kevin Martin
12 years, 6 months
School me on server hard drive management. (Hot swap, SATA channels, LVM, backup, etc.)
by linux guy
I'm building a home server that will run a MythTV backend, zoneminder and a
NAS that will stream audio content to various players and images to various
viewers, including Digikam users, as well as for general data backup from
laptops, etc.
Its based on a i5 K2500 processor in a MB with an H67 6 channel SATA
interface. (4x SATA II + 2 x SATA 3) It has 8 GB of RAM which can easily be
bumped to 16 GB. I've added a PCI 4 channel SATA I controller as well.
Both the MB and the controller support their own flavor of software RAID.
I put all this in an Obsidian 650 case outfitted with a 4 bay trayless hot
swap cage, in addition to the single hot swap slot built into the case
itself.
I also equipped it with a dual layer BD burner.
People on the MythTV user group tell me that I need to keep the Myth
recording/playback hard drive separate from other drives on the machine so
that interruptions in disk IO from other applications don't stall the
streaming of content to the myth Front Ends or writing data from the 2 HD
PVR devices. (Max 17 Mb/sec each)
Similarly, the image files should probably also have their own drive so that
when the Digikam user pulls down 200 20 MB images in a directory s/he
doesn't stall the music streaming to other devices around the house.
I have a single 500 GB 7200 RPM drive for the OS which is F15.
In summary,
i5, 8 GB RAM
2 SATA III channels
4 SATA II channels
4 SATA I channels
4 trayless hot swap bays in a cage
1 trayless hot swap bay in the case
8 non hot swap bays
1 OS hard drive
At least 1 MythTV hard drive
At least 1 images hard drive
1 audio files hard drive
more hard drives as needed.
Questions
1) Where should I physically put each drive in the machine ? Which hard
drives get hot swap and which don't.
2) Which drive should get which SATA channel ?
3) How does one mount the drives for the OS to access ? I know all about
the mount command and auto mounting, etc, but how do I reference the drives
?
For example, lets say that for some reason I pull the drive that normally
occupies /dev/sdc and reboot. The drive that was /dev/sdd will now be
/dev/sdc and thus with a "static" mount reference will mount in the wrong
mount point ????? Is LVM supposed to handle this ? It would be nice if I
could "label" a drive MythBackup and plug it into any one of the hot swap
bays and have it mount in the correct mountpoint to be used as anticipated
by its name.
SATA hotswap seems to complicate this ? I've never had hot swapping devices
before, save USB drives that I manually plugged in and mounted.
4) How should I approach data backup ? Right now I have about 250 GB of
nearly irreplaceable digital images. RAID ? Periodic backup to a (rotated,
stored off site) hard drive via the hot swap slot in the case ?
5) The drives, BIOS and controllers are all SMART capable. What use would
you make of that ?
Any and all wisdom greatly appreciated.
Thanks !
12 years, 6 months
Managing startup applications using sysctl
by Alex
Hi,
I have a few older init.d scripts that I'd like to use with my
fedora15 x86_64 system. However, they all don't start as expected and
I can't figure out why.
When running "service myprog start" it just gives a message like
"Starting myprog (via systemctl): Job failed. See system logs and
'systemctl status' for details." and no further information.
I've figured out "systemctl status myprog.service" provides some
additional details:
myprog service - SYSV: myprog is a yadda yadda yadda
Loaded: loaded (/etc/rc.d/init.d/myprog)
Active: failed since Fri, 30 Sep 2011 01:05:19 -0400; 55s ago
Process: 17956 ExecStart=/etc/rc.d/init.d/myprog start
(code=exited, status=203/EXEC)
CGroup: name=systemd:/system/myprog.service
and there is nothing in the logs from the program itself to indicate
why it didn't run. When running the command manually as it is run from
within the init script, it works just fine.
What is the proper way to debug a systemctl script?
Is there a chkconfig equivalent with systemctl or should I just
continue to use chkconfig to control which runlevels the processes run
in?
Thanks,
Alex
12 years, 6 months
Gnome-shell can't empty trash from removable (FAT) storage?
by Richard Shaw
Has anyone else noticed this or can reproduce this problem?
When I connect my android phone as a mass storage device it works as
expected other than how long it takes to be detected.
If I delete (move to trash) some files the trash bin icon stays
"empty" and the empty trash icon remains grayed out. The only
workaround I've found so far is to either show all files so I can get
to ".500-Trash" and move it to trash again which actually prompts me
to delete it or go in to a terminal and rm -rf ... it.
Thanks,
Richard
12 years, 6 months
ftp files
by Mickey
F15
what ftp app works with KDE, instead of vsftp that requires Gnome ?
12 years, 6 months