Paul's questions

I have a volume group (VG) that contains two physical volumes (PV). Several logical volumes (LV) in the VG are likely to use extents on both PVs.

Is there a way to tell which LVs occupy space on which PVs?

Varnish has the possibility to strip certain cookies from the request before deciding if a request is non-cachable because it contains cookies. Is there such a thing for Apache's mod_cache?

Consider this situation: Apache 2.2 is a reverse proxy with lots of configuration (multiple VirtualHosts, 4 backend servers, authentication based on Host and path) that I really like - because it works.

Now to speed up delivery of a Django site, I would like to cache all requests to a certain VirtualHost - except those that carry the Django session cookie (e.g. logged in staff requests). Now the request obviously Varys with cookies, e.g. a logged in user with cookie is always given a freshly rendered page while someone without cookies is given a cached response.

So Vary: cookie would be the correct cache control header to do exactly that: Always freshly render when cookies present. BUT then there is Google Analytics (GA) (and other cookie-based tracking mechanisms like eTracker) that always sets a cookie.

Can Apache ignore that GA-cookie and Vary on the Django session cookie??

I am using iperf to measure my uplink speed between my home PC behind NAT and a fully accessible root server. It does so using a TCP connection. For whatever reason it seems that iperf can only measure client to server speed, where client is the one starting the TCP connection and server is the listening part.

Unfortunately my PC is behind NAT and the TCP connection can only be initiated in this direction.

Is there another tool that does this but uses the TCP connection in both directions?

Using Duplicity I can make an incremental, efficient backup to an untrusted remote storage. This works fine and I can restore files at a specific point in time (or the next backup after that).

Well now I want to know when in the history of all the backups did a specific file change. File change timestamp and size along with the backup timestamps would be sufficient.

Any way to do that with duplicity?

EDIT: I should clarify that I want to see every change/version to that file and not only the last, e.g. when it was first introduced, when it was change. That might be at multiple backup incremental backups.

Just installed the hypervisor and it is up and running fine (in Test mode). Using the corresponding version of vSphere Client works fine too. Now sites like this show a Web Client that I just cannot see. When I visit https://my-esxi5/, the vSphere Welcome Screen does not show the Web Client link at the right side. Also, https://my-esxi5:9443/ seems to not be an open port.

Is it available at all in the free version?

This is one of the nightmare days: A virtualized server running on a Linux SW-RAID1 runs a VM that exhibits random segfaults in seemingly random codechunks.

While debugging I find that a file gives different md5sums on each and every run. Digging deeper I find this: The raw disc partitions that make up the RAID1 mirror contain 2 bit-differences and ca. 9 sectors are completely empty on one disc and filled with data on the other disc.

Obviously Linux gives back a sector from a undeterministically chosen disc of the mirror set. So sometimes the same sector is returned OK, sometimes the corrupted is given back.

The docs say:

RAID cannot and is not supposed to guard against data corruption on the media. Therefore, it doesn't make any sense either, to purposely corrupt data (using dd for example) on a disk to see how the RAID system will handle that. It is most likely (unless you corrupt the RAID superblock) that the RAID layer will never find out about the corruption, but your filesystem on the RAID device will be corrupted.

Thanks. That will help me sleep. :-/

Is there a way to have Linux at least detect this corruption by using sector checksumming or something like that? Would this be detected in a RAID5 setup? Is this the moment I wish I used ZFS or btrfs (once it becomes usable without uber-admin capabilities)?

Edit: I am not alone.

Taking over a Windows Small Business Server 2003 R2 (SBS2003) that has long been backed up using both Symantec Backup Exec and NTBackup. Now I am trying to be smart and test the restore. Of course this has been tested in the 2 years since installation, eh?

Well, restore fails with NTBackup Event IDs 8003 and 8019 being logged.

The situation: I back up System State and the complete Microsoft Information Store (in the tree under Microsoft Exchange Server -> <servername>. This completes just fine and leaves me with a backup file ca. 55GB in size and that is about right considering our Exchange usage.

Now I create a Recovery Storage Group in Exchange System-Manager parallel to Erste Speichergruppe. I add an unmounted database for the only mailbox store(?) we have (and that SBS allows).

Back in NTBackup I find the backup medium, and in there <servername>\Microsoft Information Store\Erste Speichergruppe (First Storage Group?). Three entries are visible:

• Log Files
• Postfachspeicher (<servername>) (Mailbox Store)
• Öffentlicher Informationsspeicher (Public Folders)

I select only the mailbox store and start the recover, thereby selecting an empty directory for temp files and the original <servername> as target. I select "Last Restore Set" and press OK. After 5-10s I get a "failed" status. The report reads somthing like this:

Exchange data could not be restored to <servername>\Microsoft Information Store\Erste Speichergruppe.

The Application event log shows entries 8003 (states the failed backup) and 8019 (no more information either) from source NTBackup.

How am I supposed to debug that?

Being very new to Windows software distribution for a small network (<50 clients) I was wondering how software packages like Adobe's Reader or Java are handled. I can deploy them as MSIs via group policies just fine. But what happens when the client software detects updates? What are common ways to handle this? Disable the software's autoupdate feature? Redeploy when the admin detects a new version?

Just fishing for knowledge, thanks for any hint.

I was used to having a post-commit hook trigger processing of the commit message by Trac (SVN repository). The comment (or action) was added immediately.

Now I switched to Mercurial as a VCS and Redmine as a ticketing system. Redmine does recognize information in changeset/commit messages like "refs #185" - but it takes several minutes to appear in the ticket.

Any ideas why this is? Is the routine to grab the changesets called periodically somehow?

Obviously my Apache httpd instance is doing reverse DNS (RDNS, give me the hostname for this IP address) lookups for each incoming client connection's IP address. This is bad. Especially since sometimes resolution fails with a missing PTR record - after 28secs.

Diagnostics: I added %D to my "combined" log style and looked at the response times this way: Clearly <1s for all those that are logged with their hostname and 20+s for those that get their IP logged.

This is what I tried:

• Turn off the server-status extension.
• Check that HostnameLookups Off is in the config.
• Check that mod_access is not given any hostname in an Allow/Deny rule.
• Check that reverse proxied servers follow the same rules.

What have I missed?

While looking for a 16- or 24-port gigabit ethernet switch with at least 6 VLANs and port mirroring (packet duplication to a monitoring port), I came up with three candidates (there are more): Cisco SRW2024, Netgear GS724T and HP ProCurve 1810G

I need any VLAN to be capable of at least one full GigE-communication between two ports that is not disturbed by any other traffic.

What I can't figure out is: Are these switches capable of switching multiple full-load GigE-links that sum up to 6xGigE? In the past this was known as the backplane capacity (especially when stacking multiple switches to form a giant broadcast domain) but now some don't give any information (HP in the case of the J9450A) and the others state "Nonblocking, store-and-forward switching mechanism" or even "Bandwidth: 48 Gbps".

Is it common knowledge that GigE switches handle anything you send to them as long as no single-link capacity is overloaded?