transmission-1.40 at centos 5 libcurl problem
When I tried to update my transmission (1.34) bittorent client to newest version 1.40 at centos 5 I got a configure error:
Requested ‘libcurl >= 7.16.3′ but version of libcurl is 7.15.5
and it is true… libcurl in centos 5 is now 7.15.5 and probably will not be updated in a short time. As a solution, I tried to rebuild a fedora 8 src package.
What I need was: curl-7.16.4-8.fc8.src.rpm and rpmbuild in my centos system. You will probably need nss-devel and nspr-devel so install it before running rpmbuild. Let’s rock:
rpmbuild –rebuild –define ‘dist .el5′ curl-7.16.4-8.fc8.src.rpm
after some time you should get files:
-rw-r–r– 1 root root 675076 Nov 11 10:19 curl-7.16.4-8.el5.i386.rpm
-rw-r–r– 1 root root 206491 Nov 11 10:19 curl-devel-7.16.4-8.el5.i386.rpm
of course, you can download them from my repo (click at them). Now you can easly install both:
# rpm -Uvh /usr/src/redhat/RPMS/i386/curl-*
Preparing… ########################################### [100%]
1:curl ########################################### [ 50%]
2:curl-devel ########################################### [100%]
Duplicate name exists
After starting windows xp pro, I got:
Duplicate name exists.
Solution:
If changing name for sth random (for example 376huf2e) does not help, check if the workgroup name is not same as one of computers in your network. For me - it solved the problem.
software raid 10 at linux
Today I was looking for some performace tests at linux filesystems, and I found some pages, where raid 10 at linux has been done by merging two raid 1 into raid 0, or simmilar…
Maybe some of you does not know, that linux mdadm support raid10, as you can see below:
# mdadm –create /dev/md2 –level 10 –raid-devices=4 /dev/sda5 /dev/sdb5 /dev/sdc5 /dev/sdd5
mdadm: array /dev/md2 started.
and check:
# cat /proc/mdstat
Personalities : [raid1] [raid10]
md2 : active raid10 sdd5[3] sdc5[2] sdb5[1] sda5[0]
4481920 blocks 64K chunks 2 near-copies [4/4] [UUUU]
[======>..............] resync = 32.2% (1445184/4481920) finish=0.3min speed=160576K/sec
e-mail delivery can’t be guaranteed
One of my business partners has something like this in his disclaimer:
Messages sent to and from CompanyName may be monitored to ensure compliance with internal policies and to protect our business. Emails are not secure and cannot be guaranteed to be error free. Anyone who communicates with us by email is taken to accept these risks.
In my opinion this is perfect definition of whole mailsystem based at pop3 and smtp protocol. It is doddored like whole Internet (which “be or not to be” is based at 13 global nameservers). The main problem is that is has been invented in time - when security was not considered.
Of course - we have the tls/ssl in mail comunication, but we cannot enforce the second server to use it, because of compatibility, and risk of loosing some mails.
No provider can guarantee email delivery. ISP’s have different rules about SPAM detection based on content, subject and how many of their users are reporting emails as SPAM. The best way to make sure that your specific survey invitation is not blocked as SPAM is to pre-test it with free email addresses from hotmail, yahoo etc. But as I said WITH NO GUARANTEE 
next problem is that we cannot be sure, that person who send us an e-mail is the real person, not an robot or sth. Yes - there is an SPF - but enforcing it like tls/ssl can make some serious problems to delivery. And like you can read here SPF in his simplicyty is vulnerable too… which is not a good news.
endian firewall community default root password
I do not like to read whole documentation to get the default password, it is wasting of time, so I use google to do it. But it was rather hard do find the default password for endian firewall (Yes, I’ve just installed it for testing purposes). So:
login: root
password: endian
re-add disk to mdadm array
Sometimes you can meet the situation when you lost a drive by accident from your mdadm configuration, as you can see below, two of my arrays has lost a drive:
# cat /proc/mdstat
Personalities : [raid1] [raid0]
md3 : active raid1 sdc2[0] hde2[1]
195358336 blocks [2/2] [UU]md2 : active raid0 sdc1[0] hde1[1]
97675008 blocks 64k chunks
md1 : active raid1 sda1[0]
104320 blocks [2/1] [U_]md6 : active raid1 sdb4[1] sda4[0]
958084352 blocks [2/2] [UU]md0 : active raid1 sda3[0]
14659200 blocks [2/1] [U_]
to re-add it to array just:
# mdadm /dev/md1 -a /dev/sdb1
mdadm: re-added /dev/sdb1
# mdadm /dev/md0 -a /dev/sdb3
mdadm: re-added /dev/sdb3
now - my array is alive and kicking again
# cat /proc/mdstat
Personalities : [raid1] [raid0]
md3 : active raid1 sdc2[0] hde2[1]
195358336 blocks [2/2] [UU]md2 : active raid0 sdc1[0] hde1[1]
97675008 blocks 64k chunksmd1 : active raid1 sdb1[1] sda1[0]
104320 blocks [2/2] [UU]md6 : active raid1 sdb4[1] sda4[0]
958084352 blocks [2/2] [UU]md0 : active raid1 sdb3[2] sda3[0]
14659200 blocks [2/1] [U_]
[>....................] recovery = 1.8% (272128/14659200) finish=2.6min speed=90709K/secunused devices:
create vswitch from commandline
VMware ESX supports some comandline tools, and if you would like to create a new virtual switch without a lot of clicks at vmware infrastructure client, you can easly use:
esxcfg-vswitch -a testvswitch:80
where:
-a - means add
testvswitch - is name of your virtual switch
:80 - is number of port’s you would like to have
resolution: test every disk even if its state is NEW
Some time ago, I bought six samsung 1TB drives (P/N HD103UJ), and 3 of them has been damaged (they were pretty new, and I bought them directly from distributor!)!. Two of them has read-write errors (called bad sectors ) and one does not spin up (it has S.M.A.R.T error just at system boot). Why I chose Samsung? Because this was the only one which produced 1TB with 3 plates (6 heads). Now, because I do not trust Samsung at all I bought just a newness Seagate 1TB (P/N ST31000333AS) (same 6 heads). Samsung was a trauma for me, I wonder if Seagate has same problems. I tested this two drives, and just at start I found sth strange. One of disk does not start in one of SATA ports (another disk worked fine (I swapped them). Anyway after booting up the system, I started testing:
# smartctl -A /dev/sdf |grep On
9 Power_On_Hours 0×0032 100 100 000 Old_age Always - 0
# smartctl -A /dev/sdg |grep On
9 Power_On_Hours 0×0032 100 100 000 Old_age Always - 3
Why one of disks has 3 working hours and one has 0? Maybe because they were tested by support?
now is a time for whole surface test:
# badblocks -s -v /dev/sdf
Checking blocks 0 to 976762583
Checking for bad blocks (read-only test): done
Pass completed, 0 bad blocks found.# badblocks -s -v /dev/sdg
Checking blocks 0 to 976762583
Checking for bad blocks (read-only test): done
Pass completed, 0 bad blocks found.
now - I am little bit more sure and I can use it
so maybe some performance test? because I do not have enought time to use bonnie++ I use a simple dd.
read:
# dd if=/dev/sdf of=/dev/null bs=1M count=8000
8000+0 records in
8000+0 records out
8388608000 bytes (8.4 GB) copied, 72.6436 s, 115 MB/s# dd if=/dev/sdg of=/dev/null bs=1M count=8000
8000+0 records in
8000+0 records out
8388608000 bytes (8.4 GB) copied, 69.6747 s, 120 MB/s
write:
# dd if=/dev/zero of=/dev/sdg bs=1M count=8000
8000+0 records in
8000+0 records out
8388608000 bytes (8.4 GB) copied, 70.9014 s, 118 MB/s# dd if=/dev/zero of=/dev/sdf bs=1M count=8000
8000+0 records in
8000+0 records out
8388608000 bytes (8.4 GB) copied, 71.8908 s, 117 MB/s
nice 
enabling interrupts - vmware esx 3.5i hangs at this
After making my magic usb key with vmware I was trying to install vmware esx 3.5i at HP ProLiant DL160 G5 (two quadcore processors, a lot of ram, and 4 sata 1TB disks). But with no success… The problem was that installer hanged up at “enabling interrupts” screen.
The solution I found here you should just add acpi=off to your bootloader promt, as it is said in the vmware communities:
to enable installing ESX on HP Proliant DL160 G5 you should:
1. Press Tab on very first screen of ESXi install
2. Move cursor just after first tgz filename and write “acpi=off”
3. press Enter and ESXi successfully install on DL160G5
4. Be careful because after this setting is lost and problem appears again - solution:
a) disconnect HDD, connect to computer with Win XP (Marti: or another else!), then edit boot.cfg on second partition andin second line set “kernelopt=acpi=off”
b) after starting ESX go to console (Alt-F1, see also “usupported” mode and “enable SSH in ESX” topics) type vi /bootbank/boot.cfg and set “kernelopt=acpi=off”