475 reputation
521
bio website
location
age
visits member for 2 years, 11 months
seen Nov 25 at 19:22

Apr
24
comment “Never take kernel updates” — Any truth to this?
Here's the referenced Linus story. It's an interesting look into FOSS project administration at scale: developers.slashdot.org/story/12/12/29/018234/…
Nov
19
comment How can I force a merge of all WAL files in pg_xlog back into my base “data” directory?
Are you saying that when I issue the stop command to the database server, all of the WALs in the pg_xlog directory get applied to the base data set? If so, why aren't they then deleted? The same is true for CHECKPOINT commands; I can issue one, and the WALs contained in pg_xlog shift by one sometimes (the oldest vanishes and a new one is created), but the volume stays the same, as per checkpoint_segments.
Nov
19
comment How can I force a merge of all WAL files in pg_xlog back into my base “data” directory?
Ideally, the procedure I am looking for could be run on a server post-restoration of a filesystem level backup, and would flush WAL segments shipped with the backup, as well.
Nov
19
comment How can I force a merge of all WAL files in pg_xlog back into my base “data” directory?
Your corrections are accurate. That is what we do; a filesystem-level backup, preceeded by pg_start_backup, and followed by pg_stop_backup. I know that a few WALs are necessary; at least the one closed when pg_stop_backup is called; that's why I am asking for a "near-zero" WAL volume, rather than a totally empty pg_xlog directory.
Oct
31
comment Is there a way to centrally manage Dropbox/centrally disable LAN sync for all Dropbox clients on a Windows LAN?
That's not a problem, since nobody uses the LAN sync functionality. That functionality is only useful if you have two workstations on the same LAN with your dropbox installed on both. Management is okay with it, mostly because the netadmins have a fair amount of clout.
Oct
15
comment Is it worth hiring a hacker to perform some penetration testing on my servers?
This is one of the most durable solutions yet proposed: Nessus, Splunk, and some general "wierd behavior" monitor like Zenoss when properly configured are your best friends. However, a lot of people assume they are "drop in" solutions, which they aren't. There's a very high setup overhead, and, once done, a very high security yield.
Sep
30
comment How can I block a specific type of DDoS attack?
+1 for best answer. This isn't any more "right" than making iptables rules manually, but fail2ban is much easier to admin. Just make sure you don't ban your own logins via SSH...this happened to me (forgot my password before Monday coffee...it was a bad day).
Sep
30
comment How does Windows 2008 R2 run 32-bit applications?
For more info, see en.wikipedia.org/wiki/WoW64. WOW64 does a good deal more than just translating 32-bit instructions into 64-bit ones; it's a very large and complex compatibility layer that interacts with a lot of different parts of the OS.
Sep
29
comment Why can't you mount a windows share in linux and make it writable?
If you've tried as much stuff on the Linux side as you can, I'd look at the Windows side. Create a dummy account on the windows box with promiscuous perms on the share, and have Linux connect using that account's credentials. Windows handles the administrator account differently in a lot of ways; this might be a symptom of that.
Sep
27
comment Does the Intel DX79TO motherboard support x8 devices (SAS HBAs) on PCIe x16 slots?
Verified on this particular board: both x16 slots will work with x8 cards.
Sep
20
comment Scripting postgres synchronization--is this secure?
I'd also consider (assuming you have postgres 9 or better) using pg_basebackup to synchronize your data to your development database using streaming replication. This is quicker, has no performance issues, resolves the password storage problems (by keeping the passwords in the postgres allowed users' file), and has added connection security if you're using hostssl in your pg_hba.conf file for streaming replication.
Sep
20
comment Fast Database to store 1 Billion records
Crash-proof, fast, storage-efficient, and scalable, huh? I'd start with finding one of these: en.wikipedia.org/wiki/Holy_Grail
Sep
20
comment kill -9 a postgres process
Nice answer. An addendum: if you have a backend process that absolutely will not die--not with pg_terminate_backend, not with a server stack restart, not with anything, you can kill it however you want, but make sure you have a working backup of your database. You can do that a couple of ways: you can use pg_basebackup or similar (or just rsync and pg_start\stop_backup) to back up your data directory (test the backups before continuing!), or you can use pg_dump[all] to salvage your data. Only then should you consider kill -9, or a reboot, or whatever.
Sep
20
comment Contrib directory in PostgreSQL 9.1
When I install Postgres91 and contrib from the PGDG repo on CentOS 6, the contrib files/binaries are installed into /usr/pgsql-9.1, alongside the server package's binaries.
Sep
20
comment CentOS disable filesystem check: superblock last mount time is in the future
If I could, I would. But the developers need to set the machines' dates to weird values multiple times, after kicking the systems. And it's not feasible to have a build's cleanup set the date back to a sane value because, no joke, we have a significant number of build failures that page fault the systems.
Sep
20
comment How do physical disks passed through ESXi perform vs. passing through HBAs?
I'd primarily be using SSDs for a L2ARC, but may need ZIL at some point as well. My main aim with this question was to determine if I should use RDM vs HBA Passthrough. You're right that the management overhead changes based on which I choose.
Sep
20
comment How do I force all devices on a network to register with IT before connecting, without vulnerability to Static IP-ing or MAC spoofing?
Interesting. How/with what technology?
Sep
20
comment How do I force all devices on a network to register with IT before connecting, without vulnerability to Static IP-ing or MAC spoofing?
But does that work for wirelessly-connected clients? Also, I've worked in networks that use port security before, and it's vulnerable to MAC address spoofing.
Jun
3
comment What is the max supported number of SATA devices (using cable adapters) on a Dell SAS 6/iR adapter?
By "SAS port", do you mean physical connector, or port as in "4 port internal SAS connector"?
May
29
comment Could SpinRite possibly be messing up my laptop hard drives?
No, they're pretty heterogenous. Mostly Seagate, with 2 WD (though maybe they're the same manufacturer now). Different sizes, model years, etc.