From: adam@address.invalid
Moe Trin wrote:
> On Thu, 07 Jun 2012, in the Usenet newsgroup alt.os.linux.mandriva, in
article
> , Adam wrote:
[boot schemes]
>>> It's not as if doing a disk check is an every-day affair, so
>>> having the extra delay to complete it before booting multi-user
>>> should not be a burden.
>
>> But if I have 10-15 ext2/3/4 partitions and the counter for fsck is
>> 25-35, then that would happen about 1/4-1/2 of the boots.
>
> Not for most. Most installs set the counters on all filesystems to
> the same number ('man tune2fs' and look at the -c and -i options and
> use the -l option to view the data for each partition
I see you and Dave have already discussed this. When I've used mke2fs
for ext3, it has set "maximum mount count" to a random(?) value within a
certain range, to intentionally stagger the checks.
[adam@eris ~]$ for D in a1 a3 a4 a5 a6 a7 b5 b6 b7 b8 ; do sudo tune2fs
-l /dev/sd$D ; done | grep -i "mount count"
Mount count: 19 [ext4]
Maximum mount count: 21
Mount count: 33 [ext3]
Maximum mount count: 39
Mount count: 30 [ext4]
Maximum mount count: -1
Mount count: 72 [ext4]
Maximum mount count: -1
Mount count: 68 [ext4]
Maximum mount count: -1
Mount count: 72 [ext4]
Maximum mount count: -1
Mount count: 33 [ext3]
Maximum mount count: 39
Mount count: 14 [ext3]
Maximum mount count: 29
Mount count: 6 [ext3]
Maximum mount count: 33
Mount count: 6 [ext3]
Maximum mount count: 26
[adam@eris ~]$
Somehow /dev/sda1 is the only ext4 partition with a positive number
there. 9 of those 10 partitions are mounted at every boot, and I'm
still playing with 4 distros besides my "production" one, so there must
be days when I reboot a dozen times to switch from one distro to
another. Mandriva 2011.0 stops the boot process until each e2fsck is
finished, while openSUSE 12.1 continues. When sdb5 (400 GB) is due
for checking, openSUSE gets to the "login:" prompt while fsck is still
running. On the positive side, I suppose openSUSE boots faster that
way. On the negative side, what if one of my install scripts (or my
.bashrc) uses files on sdb5?
[wireless and Ethernet]
>>> Have you discovered if the appropriate wireless module is present?
>
>> Yes, it's now present on all four (five if you include "production"),
>> and all can now attempt a wireless connection.
[..]
>> Not all succeed, though, I think because my WPA2-PSK key is a
>> character or two longer than my router can handle.
>
> "Should" be the same for all - typ0s?
Nope, because for all of them I'm using cut-and-paste from the same
file, which was modified from 'strings' output. Both the SSID and
passphrase are the maximum number of characters allowed (30 & 63). I
learned the SSID is a command-line parameter and filename in some
distros, so it can't have any characters with special meanings, like
quotes or parentheses or forward or backslashes.
[my next computer]
>> If you've been following the "get your non-UEFI hardware while you
>> can" thread here, you can see I've decided to get my next system a
>> little sooner, before only "Windows 8-ready" systems are all that's
>> available. My latest goal is to get it within the next two months.
>
> I'm sure there is going to be a way around the UEFI systems.
Yes, but so far (besides Fedora) there seems to be nothing but petitions
to OEMs. From reading the threads in this NG, I'm getting worried that
my choice is either to buy something usable NOW, or possibly wait up to
several years until things work themselves out.
>> I'm planning on an HP full-size
>> tower, quad-core, 64-bit, 6-8 GB RAM, 750+ GB HD, for $500-600.
>
> "quad-core, 64-bit" what? HP is selling both AMD and Intel, and in
> the Intel line, there's a heck of a difference in price (as well as
> performance) between an i3 and i7. At work, I'm seeing i5s and i7s
> in the servers, and Phenoms, i3s or i5s for the average workstations,
> though that's not a hard/fast rule.
As I understand things, AMD and Intel CPUs are comparable in price,
performance, and support by applications. If Intel, at least an i5.
Since I'll only be dealing with one brand, I don't think it should be
that hard to determine relative computing power within their line. It
looks to me like there's a sort of "knee" in the price/performance curve
around US$600. The bottom end is around $400, but going up to $600
gives a significant increase in performance. Going up from $600 to $800
doesn't seem to give as much of an increase.
I'm now thinking that I should buy a new tower very soon, where I can
install whatever distros I like at any time, and leave my own "distro
shootout" until after that.
[kitty news]
> Good Sam had a molar removed Friday, and is on amoxicillin and a pain
> killer (buprenorphine) twice a day for a week. Smokie is tolerating
> her "once a day" slightly better.
Have you seen the "how to give a cat a pill" piece that's floating
around the web? Today I went out to check my mail, and the grey cat
came over to me again for some head scratches. Then I went to get
something out of my car, and he went into an apartment. Not the one he
lives in, mind you, just a ground-level apartment where someone had left
the window screen open.
[ZIP codes]
>> The people who assigned the +4 codes may not have spoken to the ones
>> who actually deliver the mail.
>
> I can believe that - I suspect the codes were assigned here before
> they found that the mail boxes would be grouped rather than located
> on the individual houses.
I think the USPS would /like/ (but can't require) the owner of this
complex to install cluster boxes for everyone out by the road, instead
of the relatively small in-wall boxes in each building.
>> I heard that any building with four or more mailboxes gets its own
>> +4.
I checked the USPS website, and this building (12 apartments) has two
ZIP+4s, even though all 12 mailboxes are together.
Adam
--
Registered Linux User #536473
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
