basin's questions

My Apache httpd proxy server has an upstream proxy that only accepts HTTP CONNECT method.

According to https://httpd.apache.org/docs/2.4/mod/mod_proxy.html#proxyremote

When using https, the requests are forwarded through the remote proxy using the HTTP CONNECT method.

I added the following directive to httpd.conf:

ProxyRemote "*" "https://127.0.0.1:9051"

However, when a client sends a proxy GET request it's not converted to a CONNECT request, but is passed through unchanged.

Did I misunderstand the doc? How to make Apache convert the first proxy request to a site to a CONNECT request to the upstream proxy? Here's the desired conversation with HTTP keepalive:

browser                          Apache                 Upstream
   >> GET http://site-a.com/page1
                                >> CONNECT site-a.com:80
                                       >> GET /page1
   >> GET http://site-b.com/page1
                                >> CONNECT site-b.com:80
                                       >> GET /page1
   >> GET http://site-a.com/page2
                                (reuse existing site-a connection)
                                       >> GET /page2
   >> GET http://site-b.com/page2
                                (reuse existing site-b connection)
                                       >> GET /page2

With Options Indexes and LogLevel rewrite:trace6 I see that my rewrite rule is applied to all files in the directory being listed. It even goes inside sub-directories and looks for index.html. This is bad for performance. Is this really necessary? How to disable it? In the actual configuration I have a "prg" RewriteMap and this program receives all listed files on its STDIN.

The problem is reproducible with this minimum of configuration:

Environment:

• OpenSUSE Tumbleweed
• apache2-2.4.51-1.1.x86_64

Configs:

• /etc/apache2/conf.d/test.conf (see below the full httpd.conf)

  LoadModule rewrite_module /usr/lib64/apache2/mod_rewrite.so
  LogLevel debug rewrite:trace6
  <Directory "/srv/www/htdocs/test">
      Options Indexes FollowSymLinks
      RewriteEngine on
      RewriteRule badrule neverhappen
  </Directory>
  

Commands:

mkdir /srv/www/htdocs/test
touch /srv/www/htdocs/test/zzzzzzzzzzzzzzzzzzzzz{a..z}
systemctl restart apache2
tail -F /var/log/apache2/error_log &
curl -s -o /dev/null http://localhost/test/

Output:

[rid#556dab9a2060/initial] [perdir /srv/www/htdocs/test/] strip per-dir prefix: /srv/www/htdocs/test/ ->
[rid#556dab9a2060/initial] [perdir /srv/www/htdocs/test/] applying pattern 'badrule' to uri ''
[rid#556dab9a2060/initial] [perdir /srv/www/htdocs/test/] pass through /srv/www/htdocs/test/
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] strip per-dir prefix: /srv/www/htdocs/test/index.html -> index.html
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] applying pattern 'badrule' to uri 'index.html'
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] pass through /srv/www/htdocs/test/index.html
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] strip per-dir prefix: /srv/www/htdocs/test/index.html.var -> index.html.var
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] applying pattern 'badrule' to uri 'index.html.var'
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] pass through /srv/www/htdocs/test/index.html.var
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] strip per-dir prefix: /srv/www/htdocs/test/HEADER.html -> HEADER.html
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] applying pattern 'badrule' to uri 'HEADER.html'
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] pass through /srv/www/htdocs/test/HEADER.html
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] strip per-dir prefix: /srv/www/htdocs/test/zzzzzzzzzzzzzzzzzzzzza -> zzzzzzzzzzzzzzzzzzzzza
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] applying pattern 'badrule' to uri 'zzzzzzzzzzzzzzzzzzzzza'
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] pass through /srv/www/htdocs/test/zzzzzzzzzzzzzzzzzzzzza
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] strip per-dir prefix: /srv/www/htdocs/test/zzzzzzzzzzzzzzzzzzzzzb -> zzzzzzzzzzzzzzzzzzzzzb
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] applying pattern 'badrule' to uri 'zzzzzzzzzzzzzzzzzzzzzb'
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] pass through /srv/www/htdocs/test/zzzzzzzzzzzzzzzzzzzzzb
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] strip per-dir prefix: /srv/www/htdocs/test/zzzzzzzzzzzzzzzzzzzzzc -> zzzzzzzzzzzzzzzzzzzzzc
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] applying pattern 'badrule' to uri 'zzzzzzzzzzzzzzzzzzzzzc'
[rid#556dab9a8090/subreq] [perdir /srv/www/htdocs/test/] pass through /srv/www/htdocs/test/zzzzzzzzzzzzzzzzzzzzzc
...

Full httpd.conf:

# sudo -u nobody /sbin/httpd -d /etc/httpd-test -f /etc/httpd-test/httpd.conf -DFOREGROUND
#
# The following is not needed on SUSE
LoadModule mpm_event_module modules/mod_mpm_event.so
#LoadModule mpm_prefork_module modules/mod_mpm_prefork.so
#LoadModule mpm_worker_module modules/mod_mpm_worker.so

# The following is not needed on SUSE
LoadModule unixd_module modules/mod_unixd.so

LoadModule authz_core_module modules/mod_authz_core.so
LoadModule autoindex_module  modules/mod_autoindex.so
LoadModule rewrite_module    modules/mod_rewrite.so

ErrorLog "|/bin/cat"
LogLevel debug rewrite:trace6
PidFile /dev/shm/httpd-test.pid

ServerName localhost

Listen 8080

DocumentRoot "/srv/www/htdocs"

<Directory "/srv/www/htdocs/test">
    Require all granted
    Options Indexes FollowSymLinks
    RewriteEngine on
    RewriteRule badrule neverhappen
</Directory>

I'm logged in on a Windows Server 2012 R2 host and I need to access files on webdav servers. One is Sharepoint and the other is Apache. The Desktop Experience feature was installed and the server was rebooted.

I can map both shares as network drives from command line either with pushd or net use, however , if I try to access the shares via a UNC path it fails. Windows 7 and Windows 10 clients have no such problem.

C:\>pushd \\sharepoint.acme.com@SSL\PWA

Z:\>dir
 Volume in drive Z has no label.

 Directory of Z:\
  ...    
              18 File(s)         10 173 bytes
              46 Dir(s)  26 346 717 184 bytes free

Z:\>popd

C:\>dir \\sharepoint.acme.com@SSL\PWA
The system cannot find the path specified.

For certain reason (multiaccount) I use an alternative IMAP client (Thunderbird) with our Domino server. Recently, I noticed that if a message comes from outside of our organization its headers are encoded as LMBCS, e.g. all Cyrillic character are prefixed with 0x05 and each characters itself is encoded in windows-1251. This happens even though the sender used UTF-8-prefixed headers:

sent:
  From: =?UTF-8?B?0JjQu9GM0Y8g0JHQsNGB0LjQvQ==?= <...>
inbox:
  From: "?И?л?ь?я ?Б?а?с?и?н" <...>

There's a couple of mail robots in our organization and they somehow manage to put mail in my mailbox with UTF-8-prefixed headers.

Is there a setting to not convert the mail headers (either global or per-user)?

The problem is Thunderbird does not fully support LMBCS. When replying, it copies the headers, but replaces the characters with question marks.

My desired haproxy behavior is the following:

1. don't fail to start when failed to resolve the backend address
2. use libc to resolve addresses at start time; no DNS servers should be in config
3. use libc to resolve addresses at run time
4. always retry name resolution, if previous attempt failed
5. Don't spam my DNS server every 10 seconds. Only resolve name when needed

_

• for (1) I've added none to init-addr
• for (2) I've added libc to init-addr

Not sure what to do for (3). The resolvers section seems to support DNS only, what the hell? If I add a nameserver there and resolvers to my server string, it starts to work, but it sends a DNS request every 10 seconds. I don't want that. When I'm making a client connection I'm willing to wait for haproxy to resolve the backend using libc, synchronously.

I disconnect from network, make sure that the backend is unresolvable and restart haproxy. Then I connect to network, make sure that the backend is now resolvable and try to use a client. Haproxy still refuses to retry connecting to the backend until I restart or reload it:

Apr 25 11:34:28 il systemd[1]: Starting HAProxy Load Balancer...
Apr 25 11:34:28 il haproxy[3411]: [WARNING] 114/113428 (3411) : parsing [/etc/haproxy/haproxy.cfg:49] : 'server app1' : could not resolve address 'basinasrv', disabling server.
Apr 25 11:34:28 il haproxy[3411]: Proxy main started.
Apr 25 11:34:28 il haproxy[3411]: Proxy app started.
Apr 25 11:34:28 il haproxy[3411]: [NOTICE] 114/113428 (3411) : New worker #1 (3412) forked
Apr 25 11:34:28 il systemd[1]: Started HAProxy Load Balancer.
...
Apr 25 11:46:21 il haproxy[3412]: 127.0.0.1:45302 [25/Apr/2019:11:46:21.946] main app/<NOSRV> 0/-1/-1/-1/0 503 212 - - SC-- 1/1/0/0/0 0/0 "CONNECT serverfault.com:443 HTTP/1.1"

_

defaults
    # never fail on address resolution
    default-server init-addr libc,none
    #default-server init-addr last,libc,none
    option log-health-checks
    option httplog

global
    ca-base /etc/ssl/certs
    maxconn     20000
    log         /dev-log local0 debug
    user        haproxy
    chroot      /usr/share/haproxy
    pidfile     /run/haproxy.pid
    daemon

frontend  main
    bind :3129
    mode                 http
    log                  global
    maxconn              8000
    timeout              client  30s
    default_backend             app

backend app
    log                  global
    mode        http
    reqadd      Proxy-Authorization:\ Basic\ xxxxxx
    timeout     connect 5s
    timeout     server  30s
    timeout     queue   30s
    server  app1 basinsrv:443 check ssl verify required ca-file ca-certificates.crt

I have a root domain I have control of and a set of subdomains, but other people are responsible for them.

I want to add a CAA protection to my root domain, but I don't want to restrict the subdomain users from using certification authorities of their choice.

Unfortunately, subdomains inherit the issue tag of a parent domain. Is there a technical possibility to allow any authority issue a certificate to a specific subdomain? An empty string means "no one".

My workstation is connected to one of the corporate ethernet switches. There are multiple VLANs. ipv4 broadcast and ipv6 FF02::1 only work within current VLAN.

I have enabled wake on lan and waking it with a magic packet from another workstation, which is in another VLAN/subnet. Therefore I cannot use a broadcast address. Instead I have to use a unicast address.

However, this only works if I put an intermediate 5-port domestic switch between my workstation and the corporate switch. Can somebody explain that?

It is obviously not a timeout, because I send the packet right after I put my workstation to sleep. It looks like that side somehow notices my actions and instantly erases the ARP and ipv6 neighbor cache. With a domestic switch WOL works, at least for several minutes.

My network card is old and does not support ARP offload (I never saw one).

Upd:

I was able to simulate a sleeping workstation by running this command on Linux:

iptables -I OUTPUT -j DROP

...and then launched tcpdump. Initially a lot of packages come. If I replug the cable between my switch and the PC, packages are still coming, but if I replug the cable between my switch and the corporate switch, they stop coming. There are rare CDPv2 packets.

I think that when I put my workstation to sleep, a short disconnect occurs, possibly due to the card lowering the network speed. My card has no "WOL speed" setting.

I have RHEL 6.8 installed on a fake Intel RAID1 (Intel Matrix). It's managed by dmraid, consists of sda and sdb and the device name is /dev/mapper/ddf1_4c5349202020202080862682000000004711471100000a28p1

Now I want to write something else to sdb. How to do that without reboot? On mdadm-managed arrays it's easy:

mdadm --fail /dev/md126 /dev/sdb
mdadm --remove /dev/md126 sdb

I have 2 iSCSI fileio backstores.

I try to Determine whether shared disks can be used as tiebreaker disks . For that I run on the iSCSI initiator a tool from IBM to release the reservation:

[root@softiwarp-1 ~]# /media/distr-ro/db2/10.5/linuxx64/v10.5fp7_linuxx64_universal_fixpack/universal/./db2/linuxamd64/install/tb_break -u -t SCSI WWID=`scsi_id -g -u /dev/sdd`
Initializing SCSI tie-breaker (WWID=36001405bb9264c9e8c145429dbc76047)
Releasing tie-breaker (WWID=36001405bb9264c9e8c145429dbc76047)
tb_release status 0 (errno=0)
[root@softiwarp-1 ~]# /media/distr-ro/db2/10.5/linuxx64/v10.5fp7_linuxx64_universal_fixpack/universal/./db2/linuxamd64/install/tb_break -u -t SCSI WWID=`scsi_id -g -u /dev/sde`
Initializing SCSI tie-breaker (WWID=36001405ed3a0a03628a4108bc3e24a82)
Releasing tie-breaker (WWID=36001405ed3a0a03628a4108bc3e24a82)
tb_release status -1 (errno=25)

For one of the disks the command succeeds. The other fails and on the target in dmesg I see

Received legacy SPC-2 RESERVE/RELEASE while active SPC-3 registrations exist, returning RESERVATION_CONFLICT

What does it mean?

/iscsi/iqn.20...ample:t1/tpg1> ls /
o- / ......................................................................................................................... [...]
  o- backstores .............................................................................................................. [...]
  | o- block .................................................................................................. [Storage Objects: 0]
  | o- fileio ................................................................................................. [Storage Objects: 2]
  | | o- shareddata ................................... [/home/libvirt/images/default/test_iSCSI.img (20.0GiB) write-back activated]
  | | o- shareddata_tb .............................. [/home/libvirt/images/default/test_iSCSI2.img (100.0MiB) write-back activated]
  | o- pscsi .................................................................................................. [Storage Objects: 0]
  | o- ramdisk ................................................................................................ [Storage Objects: 0]
  o- iscsi ............................................................................................................ [Targets: 1]
  | o- iqn.2014-08.com.example:t1 ........................................................................................ [TPGs: 1]
  |   o- tpg1 ............................................................................................... [no-gen-acls, no-auth]
  |     o- acls .......................................................................................................... [ACLs: 1]
  |     | o- iqn.2014-08.com.example:client ....................................................................... [Mapped LUNs: 2]
  |     |   o- mapped_lun0 ........................................................................... [lun0 fileio/shareddata (rw)]
  |     |   o- mapped_lun1 ........................................................................ [lun1 fileio/shareddata_tb (rw)]
  |     o- luns .......................................................................................................... [LUNs: 2]
  |     | o- lun0 ................................................ [fileio/shareddata (/home/libvirt/images/default/test_iSCSI.img)]
  |     | o- lun1 ............................................ [fileio/shareddata_tb (/home/libvirt/images/default/test_iSCSI2.img)]
  |     o- portals .................................................................................................... [Portals: 1]
  |       o- [::0]:3260 ....................................................................................................... [OK]
  o- loopback ......................................................................................................... [Targets: 0]

Our server Supermicro SYS-6016T-U often crashes due to https://bugs.centos.org/view.php?id=10642 .

Since it's unable to write to disk, crash info isn't saved. I also cannot login using either IPMI kvm console or ssh to run dmesg.

How to configure it to save its errors to some network location? (actually, NetworkManager crashes too, so the network card may lose its config) Or maybe the ipmi device has some buffer for that?

I made a mistake and reserved 100% disk space for software raid. Not sure I can use kdump now

I'm sharing a windows folder using both nfs and a windows share. The windows share has read permissions for Everyone. The nfs share has write permissions (with "Allow root access") for a specific host.

I created a couple of files and folders with different owners and groups including root and 4294967294 and unix permissions 777, but a windows client cannot read them.

drwxrwxrwx 2 distuser 4294967294   64 Apr 20 09:47 /mnt
drwxrwxrwx 2 distuser 4294967294 4096 Apr 20 09:57 /mnt/distr

I can list the contents of /mnt, but not /mnt/distr from a windows client.

After creating a directory using nfs, the security tab shows explicit permissions for: SYSTEM, Administrators, and some unresolvable SIDs.

Can I map the traditional unix scope "others" to NTFS "Everyone"?

Currently I have the following:

C:\pub>icacls after
after S-1-5-88-3-504:(N)
      S-1-5-88-1-0:(F)
      S-1-5-88-2-0:(RX,W,DC)
      S-1-5-88-4:(Rc,S,REA,RA)
      NT AUTHORITY\SYSTEM:(F)
      BUILTIN\Administrators:(F)

S-1-5-88-4 is the SID currently mapped to others.

Is it possible to stop listening on iSCSI server port 3260 without deleting the portal from the configuration? I tried to stop all services I know, but it's still running.

[root@okvmh1 ~]# systemctl stop iscsid
[root@okvmh1 ~]# systemctl stop iscsi
[root@okvmh1 ~]# systemctl stop target
[root@okvmh1 ~]# netstat -tln | grep 3260
tcp6       0      0 :::3260                 :::*                    LISTEN
[root@okvmh1 ~]# iscsiadm --mode discovery --type sendtargets --portal localhost
[::1]:3260,1 iqn.2014-08.com.example:t1

I want to transfer block device contents over the network directly into a qcow2 image. I started with a small simulation:

[root@okvmh1 default]# dd if=/dev/zero bs=1M count=100 | qemu-img convert -p -f raw -O qcow2 /dev/stdin aaa.qcow2
qemu-img: Could not open '/dev/stdin': Could not refresh total sector count: Operation not permitted
qemu-img: Could not open '/dev/stdin'

It seems that qemu-img does not know the desired image size. In VirtualBox I would do:

<net-command> | VBoxManage convertfromraw stdin aaa.vdi $size --format VDI

But qemu-img convert does not accept the size parameter.

Need to provide an anonymous readonly access to an svn repository

AFAIK, haproxy cannot auth on backend by itself.

There's a working configuration for haproxy with basic auth. How to migrate it to NTLM? Main problem is we cannot simply add a static header because of the challenge-response mechanism.

frontend svn_f
  bind *:8000
  mode http
  default_backend svn_b
  acl valid_method method GET HEAD OPTIONS PROPFIND REPORT
  acl valid_path path_beg /svn/myrepo
  http-request deny unless valid_method valid_path

backend svn_b
  reqadd      Authorization:\ Basic\ xxxxxxxxxxxxxxxxxxxx
  server svn_server 192.168.2.1:443 ssl verify none

Maybe Apache can do this? Tried this, but it does not work:

SSLProxyEngine on
<VirtualHost *:8000>
KeepAlive on
ProxyPass / https://user:[email protected]/ connectiontimeout=5 timeout=30 keepalive=On
ProxyPassReverse / https://192.168.2.1/
</VirtualHost>

Have IBM db2 on Rhel on Hyper-V on Windows Server. The *.vhdx file is on CSV. e.g. C:\ClusterStorage\volume1\hyper-v\oktest-lb1\oktest-lb1.vhdx.

db2 uses O_DIRECT flag (uncached i/o) for its database files, and a small block size 8k or less for read/write operations.

Somehow this direct flag is propagated through the whole stack to the host system. The host cache is not used.

The network lag kills performance of such small operations. tested with dd with oflag=direct / iflag=direct

• bs=8k: 12Mbytes/s
• bs=64k: 100Mbytes/s

Need a workaround.

I want to redirect all local mail by specifying an address in virtual_alias_maps file.

It is possible to match recipients with regexp, but it's too complex. They can be:

• no domain (root etc.)
• root@localhost
• root@myserver

Here: http://forums.debian.net/viewtopic.php?f=16&t=107671 they try to match /.+@.+\.localdomain/ , which of course doesn't work.

Postfix already has $mydestination config parameter. Can I match with it somehow?

Upd: tried local_transport = virtual: , but virtual cannot forward.

Perhaps a global .forward file will help?

From the man page of hwclock:

This mode (we'll call it "11 minute mode") is off until something turns it on. The ntp daemon xntpd is one thing that turns it on. You can

Ok, I have a virtualbox linux guest without ntp client. There's a 2 minute difference:

$ date
Tue Mar 11 18:09:54 MSK 2014
$ hwclock 
Tue 11 Mar 2014 06:08:23 PM MSK  -0.409965 seconds

How to enable 11 minute mode?

$ adjtimex --print
       status: 64

64 means "disabled", right? http://manpages.ubuntu.com/manpages/karmic/man8/adjtimex.8.html

64 clock unsynchronized

I thought this will enable the mode

$ sudo adjtimex --status 0

But it didn't. Status is still 64

I'm connecting from Windows 7 Ent to Windows Server 2008 R2. I followed several guides to setup audio redirection. Playback works, but no recording.

• "Record from this computer" checked on client
  

• microphone on client works

• "Windows Audio" service on server running
  

Now I have 2 options:

1. uncheck 2 audio-related options in RDP-Tcp Properties

2. allow audio redirection in group policy. This makes the above options grey.

I tried both options

• set "fDisableAudioCapture" = 0. It's already set to zero by unchecking in "RDP-Tcp Properties" dialog
  

• log off and log in

• open "Recording devices" and check "Show Disabled Devices" and "Show Disconnected Devices"

No audio devices are installed

How to drop incoming negative DNS responses in Linux? (I want a timeout instead) I'm thinking of creating an iptables rule:

iptables -I INPUT -p udp --sport 53 -m u32 ...

From wireshark:

000E start of IP Packet (fixed, at first, I wrote 0010)
0022 start of UDP Packet
002A start of DNS message
002C  Flags
        .... .... .... 0000 = Reply code: No error (0)
        .... .... .... 0011 = Reply code: No such name (3)

I think the offset for -m u32 is counted from the ip header. Also subtracting 2 to get a 32bit.

$ echo $(( 0x002C - 0x000E - 2 ))
28

The mask for the last 4 bits is 0xF
So the rule should be something like this:

iptables -I INPUT -p udp --sport 53 -m u32 --u32 "28&0xF=3" -j DROP

For logrotate I have a cron task that runs hourly.
In logrotate.conf there's the directive "daily". But the log files are rotated hourly, not daily.
Why does it happen? What's this directive for? I thought that these time-related directives are for rotating more seldom than the tool executed.

$ cat /opt/etc/logrotate.conf
compress
/opt/var/log/dir300.log {
        daily
        rotate 29
        postrotate
                killall -HUP syslog-ng
        endscript
}

.

$ grep logrotate /opt/etc/crontab 
50 */1 * * * admin /opt/sbin/logrotate -f /opt/etc/logrotate.conf &>/dev/null

$ ls -l dir300.log*
-rw------- 1 admin root 159856 2013-01-08 11:29 dir300.log
-rw------- 1 admin root   6838 2013-01-08 01:50 dir300.log.10.gz
-rw------- 1 admin root     20 2013-01-08 00:50 dir300.log.11.gz
-rw------- 1 admin root    592 2013-01-07 23:50 dir300.log.12.gz
-rw------- 1 admin root    124 2013-01-07 22:50 dir300.log.13.gz
-rw------- 1 admin root     20 2013-01-07 21:50 dir300.log.14.gz
-rw------- 1 admin root     20 2013-01-07 20:50 dir300.log.15.gz
-rw------- 1 admin root    107 2013-01-07 19:50 dir300.log.16.gz
-rw------- 1 admin root     20 2013-01-07 18:50 dir300.log.17.gz
-rw------- 1 admin root     20 2013-01-07 17:50 dir300.log.18.gz
-rw------- 1 admin root     20 2013-01-07 16:50 dir300.log.19.gz
-rw------- 1 admin root  14158 2013-01-08 10:50 dir300.log.1.gz
-rw------- 1 admin root     20 2013-01-07 15:50 dir300.log.20.gz
-rw------- 1 admin root     20 2013-01-07 14:50 dir300.log.21.gz
-rw------- 1 admin root    133 2013-01-07 13:50 dir300.log.22.gz
-rw------- 1 admin root     20 2013-01-07 12:50 dir300.log.23.gz
-rw------- 1 admin root    279 2013-01-07 11:50 dir300.log.24.gz
-rw------- 1 admin root     20 2013-01-07 10:50 dir300.log.25.gz
-rw------- 1 admin root     20 2013-01-07 09:50 dir300.log.26.gz
-rw------- 1 admin root     20 2013-01-07 08:50 dir300.log.27.gz
-rw------- 1 admin root     20 2013-01-07 07:50 dir300.log.28.gz
-rw------- 1 admin root     20 2013-01-07 06:50 dir300.log.29.gz
-rw------- 1 admin root  14236 2013-01-08 09:50 dir300.log.2.gz
-rw------- 1 admin root  14135 2013-01-08 08:50 dir300.log.3.gz
-rw------- 1 admin root  14349 2013-01-08 07:50 dir300.log.4.gz
-rw------- 1 admin root  14396 2013-01-08 06:50 dir300.log.5.gz
-rw------- 1 admin root  14268 2013-01-08 05:50 dir300.log.6.gz
-rw------- 1 admin root  14148 2013-01-08 04:50 dir300.log.7.gz
-rw------- 1 admin root  14347 2013-01-08 03:50 dir300.log.8.gz
-rw------- 1 admin root  14225 2013-01-08 02:50 dir300.log.9.gz

.

$ ipkg list logrotate
logrotate - 3.7.5-2 - Rotates, compresses, removes and mails system log files.