Re: [gentoo-amd64] Re: grub and maximum kernel file size

[Branko Badrljica]

Duncan wrote:

>
>> Duncan wrote:
>>
>>
>>
>>> Did you try md-mod,start_dirty_degraded=1 (AFAIK this applies to
>>> RAID-4/5/6 only)?
>>>
>>>
>> No
>>
>
> Hmm, and you said RAID-5, right?
>
>
Yes. I tried now to boot without initramfs and it works- with and 
without that "md-mod,start_dirty_degraded=1" option.
But this is now. Will it continue to do so, remains to be seen. But no 
matter what, initramfs remains a keeper for me.

BTW: Where is this mod-module name way of setting inbuilt module 
variables documented ?
I have uvesafb built-in as part of kernel and had to check the sources ( 
documentation was out of date at some point ) to coem up with command 
line setting, like "uvesafb.mode_option=1280x1024-16@60 uvesafb.mtrr=3".

Could I also write "md.start_dirty_degraded=1" ?



> I think one of the reasons I may have never had to use that option on
> RAID-6 is that it has two-way parity, so chances are either one or the
> other will still be valid -- even if it crashes in the middle of a write,
> one stripe-write might be invalid, but as long as I wasn't already
> running degraded, the state on the others should be clean, either clean-
> state from before the write, or clean-state from after the write, thus
> ensuring data integrity.  Naturally, due to hardware cache buffers, etc,
> this isn't entirely guaranteed (unless the write buffers are disabled),
> but the chances are far higher of a clean state than they are with RAID-5
> with its single-way parity.
>
Yes, but RAID-6 isn't simple expansion of RAID-5 concept.

It has two parity chunks-P and Q. While P is simple XOR parity from 
RAID-5, Q is substantially compute intensive.
Further, while Q generation is "expensive", CPU can utilise its SSE 
artillery and soften the perceived extra load.

But when CPU has to actually use Q parity for data regeneration, things 
come to a crawl. That part is quite intensive and SSE is useless there, 
so best CPU can do is use tabelaric approach and regenerate data 
byte-by-byte.

So SW RAID-6 has come to be a bad idea, at least for me and my 
applications. HW RAID-6 with good HW is another story altogether, though.

With SW, you pay extra drive, small extra power consumption, some extra 
cycles for Q generation ( not quite inconsequiential) and BIG penalty 
for regeneration - so big that usually "doing bussines as usual" while 
in such degraded mode is out of the question.

So after some experience with SW and HW RAID-6 i opted for RAID-5 with 
cold spare and regular backups for pretty much everything.

> (FWIW I do run without UPS. Unfortunately the reality is that a proper
> UPS for this level of machine, at least back when I was running CRTs,
> would cost near what the machine itself did, and given that I was already
> buying near the top of my budget range that put a UPS out of
> consideration.  Since I switched to LCD I've been considering it again
> and would likely have one by now had the economy and therefore my budget
> not tanked.)
>
I did run for some time all of my machines through UPS only to find this 
costs too much.

So I switched to DC. I rectified all three 220VAC phases to  260-300V DC 
and backed this up by 20x 100Ah-12V lead-acid akus in series. Even 
without akus it feels pretty much close to having UPS. When interruption 
happens, 90% of the time it affects only one phase, so remaining ones 
take over and I don't feel a thing...

For my remote location I can't have 20 akus there, so I am working on 
PSU modification which would allow me to work on two...

> That's what my "rootbak" partition is for. =:^)
>
Yes. And I use same thing on my workstation. Still, that doesn't quite 
match robustness of initrd.
With initrd, if you can get to the phase where grub takes over, you can 
pretty much boot. Even if your kernel can't work with root drive for 
whatever reason ( missing driver for SATA,AHCI, ext4 ...)

> Well, you /could/ burn a CD with a kernel built with the proper options,
> but you are correct, booting from a CD is indeed a hassle.
>
On Gentoo, how to do this is far from obvious.

> Actually, I've seen mods for *-WRT class routers and software (OpenWRT,
> DD-WRT, Tomato, etc) that do pretty much the same thing, solder a USB
> based flash module to the thing and expand their storage and flexibility
> by an order or two of magnitude!  Run a server, etc, directly off your
> router!  They can be made detachable, or not, as desired.
>
>
Cool. If you have a link to such a thing, it would be apppreciated ;o)
> But if I were doing it over, as I said earlier, I'd definitely create
> separate RAID-6s instead of the single partitioned RAID-6, and use two
> rootbaks so three root filesystem images total, one on each of three
> separate RAID-6s, so if one failed to come up, I'd just change the
> command line as necessary to bring up the second, and if it failed, the
> third.
>

Not every application has plenty of storage space.

> I'd also split the RAID-1, either creating two single-disk RAID-1s or
> more likely, four separate /boot images direct on the individual spindles.
>
>
This is what I do now. Four way RAID-1 for boot is exactly that.
GRUB sees it as four separate drives with ext2 that can be mounted 
exactly the same as any other drive, so I can boot from any of them.
When kernel takes over, it assembles them into an md array, so you can 
mount /boot and update all four drives with exactly the same data, when 
needed...