my poor, old Proliant DL 380 G5 room heater

General support questions
lightman47
Posts: 764
Joined: 2014/05/21 20:16:00
Location: Central New York, USA

my poor, old Proliant DL 380 G5 room heater

Post by lightman47 » 2018/12/05 17:27:41

While it graciously accepted 7.6 without complaint, it didn't like it. After the reboot, I was greeted with it being locked (dracut errors). The next reboot made it to Plymouth screen message and subsequent reboots didn't go much better. I'm not consumed with the idea of forcing it, even if it were possible, and didn't even dig into the logs. I figured dracut complains were enough of a warning. I've decided to leave it a 862.14.4 or whatever the last 7.5 update was. I've even set the default boot to that kernel and run grub2-mkconfig to make it 'permanent'. (The machine is more "toy" than anything).

In my automatic shared backup/update script, I have coded out this machine from the updates. However, this blocks the machine from getting updates that are NOT release dependent. My question is then how can I have it run update, but NOT upgrade itself to 7.6? (Or is this a pipe dream).

no rush - I've got plenty of work ahead of me to get the GUI back on that 862 kernel. The 7.6 messed up other kernels on a number of my machines - fingers crossed I won't need them until they roll off the list.

Thank you again.
"The more I learn, the behinder I get"

User avatar
TrevorH
Forum Moderator
Posts: 24052
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: my poor, old Proliant DL 380 G5 room heater

Post by TrevorH » 2018/12/05 17:36:10

You could just add

exclude=kernel-3.10.0-957*

to both base and updates repo files.

I'd suggest booting without rhgb quiet in order to see all kernel messages if you're interested in tracking down _why_ it happens.
CentOS 5 died in March 2017 - migrate NOW!
Full time Geek, part time moderator. Use the FAQ Luke

lightman47
Posts: 764
Joined: 2014/05/21 20:16:00
Location: Central New York, USA

Re: my poor, old Proliant DL 380 G5 room heater

Post by lightman47 » 2018/12/05 18:58:20

OK - added exclude=kernel-3.10.0-957* to both sections of CentOS-Base.repo. Not knowing how versions are incremented, will 957 cover me for a long time?

{I usually hit [Esc] or arrow-key when I want to see boot.log being built.}

Still wrestling with the machine after a couple 'yum reinstall \*.\*' under two previous kernels (so they all get replaced). Still get the generic VGA "X" cursor - and no further. At least I can ssh in instead of running upstairs as the basic system is running. Some how 7.6 seems to have clobbered Gnome in all previous kernels. I needed a good project anyway.

Thank you.
jje
"The more I learn, the behinder I get"

User avatar
TrevorH
Forum Moderator
Posts: 24052
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: my poor, old Proliant DL 380 G5 room heater

Post by TrevorH » 2018/12/05 23:35:39

The usual convention is that each point release gets a new xxx in 3.10.0-xxx and then each update after that within a point release will increment the numbers following the -xxx.

The only time I've seen that convention broken was on CentOS 5 after the final point release which gave us 2.6.32-398 and then the next kernel updates, rather than giving us 2.6.32-398.1.1 etc, we got 2.6.32-400 then all the way up to 419.
CentOS 5 died in March 2017 - migrate NOW!
Full time Geek, part time moderator. Use the FAQ Luke

lightman47
Posts: 764
Joined: 2014/05/21 20:16:00
Location: Central New York, USA

Re: my poor, old Proliant DL 380 G5 room heater

Post by lightman47 » 2018/12/06 13:26:36

I wonder then if trimming "exclude=kernel-3.10.0-957" to "exclude=kernel-3.10.0-9" in the two settings will prevent updates for the foreseeable future. Or do you know if the .repo files permit wildcards (link "exclude=kernel-3.10.0-9??")?
"The more I learn, the behinder I get"

User avatar
TrevorH
Forum Moderator
Posts: 24052
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: my poor, old Proliant DL 380 G5 room heater

Post by TrevorH » 2018/12/06 16:17:43

Since the last jump was from 862 to 957, I would suspect that 7.7 will have a kernel way up in the 1000+ range.

You should probably try to debug whatever the problem is and report it so that it gets fixed rather than excluding updates. Sooner or later there will be a critical kernel security patch that you need...
CentOS 5 died in March 2017 - migrate NOW!
Full time Geek, part time moderator. Use the FAQ Luke

lightman47
Posts: 764
Joined: 2014/05/21 20:16:00
Location: Central New York, USA

Re: my poor, old Proliant DL 380 G5 room heater

Post by lightman47 » 2018/12/06 17:09:38

Oh, gee - my thought train had actually been heading toward fixing 862 video that got broke in the upgrade rather than getting 957 working. Symptoms with 7.6 seem to point to my P400 controller with RAID 1+0 and the 'hpsa' commands in the startup. Even my 7.6 USB key seems to choke on it (I tried to re-install 7.6). If I omit the 'hpsa' commands, the key boots into the install, but there are no drives. If I include the 'hpsa's in the boot, the key totally "loses it" and locks up after filling the screen with tons of error messages - rarely the same messages twice. Incidentally, this is also how the 7.6 installation behaves. The update went smoothly. but then there was the reboot and it all broke loose!

{By the way, the bug that needs reporting is that the 7.6 has somehow 'broken' previous kernels on a few of my machines. It seems to have to do with video Nvidia, and not Nvidia alike.}

I'd thought that if I couldn't get 862 video fixed, I try a re-install of 7.5(1804) which I still have. That failing, my next adventure would be to un-install the RAID in the BIOS, then try a clean install of 7.6. And, if that failed, 7.5 with no RAID.

As mentioned, this machine is just a 'toy' with which to learn. When it does run - one day a week - it only runs BOINC/SETI packets, backup, and update. It was a Christmas present from my son several years ago - that's too expensive to run any length of time.

:D
"The more I learn, the behinder I get"

User avatar
TrevorH
Forum Moderator
Posts: 24052
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: my poor, old Proliant DL 380 G5 room heater

Post by TrevorH » 2018/12/06 17:20:47

I do not think hpsa is intended for use with those old controllers. That's why you have to tell it to allow any controller, even unsupported ones. You should probably look at using kmod-cciss instead.
CentOS 5 died in March 2017 - migrate NOW!
Full time Geek, part time moderator. Use the FAQ Luke

lightman47
Posts: 764
Joined: 2014/05/21 20:16:00
Location: Central New York, USA

Re: my poor, old Proliant DL 380 G5 room heater

Post by lightman47 » 2018/12/06 17:31:17

I shall try exactly that. I'd seen you recommend it before but decided not to break what seemed to be working. Heh.


Currently pouring through it's 862 logs trying to find what went wrong with the video (that seems to be ATI Radeon - I didn't know that).

Thank you.
"The more I learn, the behinder I get"

lightman47
Posts: 764
Joined: 2014/05/21 20:16:00
Location: Central New York, USA

Re: my poor, old Proliant DL 380 G5 room heater

Post by lightman47 » 2018/12/06 19:00:35

haha - Not so much!

In 7.5 (862) I
yum installed kmod-cciss
edited out the two 'hpsa' commands in /etc/default/grub
rebuilt grub.cfg via 'grub2-mkconfig -o /boot/grub2/grub.cfg
reboot

Now even 7.5 complains in dracut (can't seem to get pastebin to accept the snapshot but will try here)

Image

OK - apparently not!

many dozens of dracut-initque[524]: warning: dracut-initqueue timeout - starting timeout scripts
then:
dracut-initque[524]: warning: Could not boot.
dracut-initque[524]: warning: /dev/centos/root does not exist
dracut-initque[524]: warning: /dev/centoes/swap doeas not exist
dracut-initque[524]: warning: /dev//mapper/centos-root does not exist
Starting Dracut Emergency Shell...
Warning: /dev/centos/root does not exist
Warning: /dev/centos/swap does not exist
Warning: / dev/mapper/centos-root does not exist

Generating "/run"initramfs/rdsosreport.txt"
{poop - you probably want that, don't you?! I'll have to hand-type whatever's in it)

Entering emergency mode. (The usual after that about Exit & saving the report to a USB stick ...
---
USB stick lights, but can't seem to find it from the dracut prompt. I did 'cat' the report which was lengthy. Getting it to the USB will be a trick.
---
All that said - this all is NO EMERGENCY/no rush.

I can always "break in" (I think) with a rescue mode and put back the 'hpsa' lines to get a boot up from which I can yum erase the cciss kmod. If not, oh well. It was a learning machine.

Thank you.
"The more I learn, the behinder I get"

Post Reply