May was a busy month on the quest. I moved to a new web server and my Windows 7 desktop got a fresh new OS installation. But not everything was new. May is a big month for birthdays in my family, including my own. At 11, this month had the highest post count so far this year..
A comment got me interested (again) in using my Synology as a Time Machine backup destination. I’m still waiting to see how it handles some low disk situations. I found I couldn’t rush the testing. It appears Time Machine complains if it doesn’t have enough backup history and it runs out of space. It simply says to add disk. So I figure after running it a month I’ll start using up space again. I think it will work OK. I have had the occasional error that seems more like a regular time machine error than Synology specific. The next backup after the error has always worked.
I rebuilt my Windows 7 desktop early in the month. Things have been running nice and quick. I haven’t been installing software until I actually need it so there’s a lot less cruft on if.
I also went through a overall IT slimming down and reviewed what I was actually using. This was a good opportunity to clean up the “What I Use” pages on this site. I deleted the current “What I Use” sections and updated them with blog posts. The first was the iPad apps I use. There’s already been a change to that list (I updated the article). Mr. Reeder has replaced NetNewsWire as my RSS Reader on my iPad. Since I moved to a new web server it was a good time to document what I use for my web server. Then I wrapped up what I use within the house in Trail Log #71, which ran to about 2,000 words. (I need to cut back what I use even more.)
Web Server Move & Changes & Frustrations
I moved this site to a new web server this month. I’m still with Linode and it’s still a 512 MB VPS. But I switched to Debian 6 (“Squeeze”). I also made some configuration changes to Apache and PHP on the server. The biggest changes were the switch to MPM-Worker for Apache and mod_FCGID for PHP. Figuring out the configuration was a bit of a hassle but it’s been solid since the move.
As luck (or Murphy’s law) would dictate there were some problems with my Linode server and connectivity after the move. I’d been running the new VPS for about 6 weeks without having a problem. Then after the move there was a network issue for a couple hours during the evening of the 23rd. The server was fine, but inaccessible or experiencing timeouts. Then there was a planned software upgrade the evening of the 26th which required a reboot. While they weren’t specific it sounded like a security update, maybe to the Xen software. Then on Wednesday there was another reboot due to emergency maintenance.
To their credit Linode alerted me and had tickets open for the problems as quickly as my own alerting system (with 5 minute checks) alerted me. And when I logged on to the console there was a banner message telling me there was a known issue with my Linode and no need for me to reboot or take other action. They also explained the network issue (it was self-inflicted by a change) and they documented steps to prevent a re-occurrence. The server issue is resolved but no specific root cause yet.
Since things were going so well I decided to move ahead and try other new things. I started to dig into website performance strategies. I added APC (Alternative PHP Cache) as an opcode cache. I also switched from WP-Supercache to W3 Total Cache since W3tc can use APC. I did some quick tests and found that W3tc and WPsc provided similar performance and features when using disk based page caching (make sense since they just create, not serve the static page). I like WP-Supercache a little better because it provides more visibility into what it’s doing. If I don’t use APC in conjunction with W3tc, or it provides minimal benefit, I may go back to WP-Supercache.
During the move I caused a self-inflicted Apache problem due to a syntax error in a logrotate file. I didn’t notice that until I saw that the logs were being rotated at the default settings, not my settings. That led me to continue squashing bugs and I went through the Apache error logs and start eliminating those frequent but inconsequential errors and warnings. I finally squashed Apache’s “file does not exist error” which had been filling the log but not affecting my site.
I just removed Google Analytics from my website, switching to the open source Piwik which I’ve been running for a few weeks.
I also removed the tweet and google+ buttons from each post. I never use them myself and I don’t know anyone who does. In my performance analysis they came up as being sub-optimal and targets for optimization. Now they have no impact on performance. Plus it’s creepy that those buttons can be used for tracking since I’m giving those companies carte blanche to put anything in the script that’s run from their server to display the button.
News That Interests Me
Comcast is rethinking there 250 GB data cap. The cynic in me says they’re looking for a way to collect more money from their customers. They wordsmith their press release too much, using words like “public internet”. As for my own usage, the only times I ever bumped against the cap was when I refreshed my backups or was testing a new offsite backup service. In April I experimented and did almost all my “TV” watching through streaming video. Everything else was normal. I don’t have any kids and I don’t consider my TV/Movie viewing above normal. I hit 190MB of usage that month. The plans they’re testing seem reasonable, but I still don’t trust them. The data cap as a network management tool doesn’t hold water. I can go over the cap by only using my connection in the dead of night (and for backups that’s when I use it) and not affect their network. Or I can watch video during prime time when everyone else is, saturating their network, and stay well under the cap. Just admit that data cap is a way to get people who use their service more to pay more than people who use it less. This is especially true if their own Xfinity video on demand service, delivered over IP, doesn’t count against the data cap. No doubt that’s the reason they use the term “public internet”. Their Xfinity service never hits their definition of the “public internet”, even though it’s using the same infrastructure they use to provide ISP service.
When I read the headline that Google Chrome was “the most used browser” my initial reaction was “BS” before even reading the story. While the headline is catchy, trying to crown a winner is pointless. One analytics firm made the claim based websites that use their tracking code. While Microsoft is hardly unbiased, they rightly point out that Chrome pre-fetches web resources which may inflate stats. I don’t doubt the trend (Chrome up, IE down) but they’re not enough accuracy to justify the headline. As for The OS Quest in May (as reported by Piwik) – Google Chrome: 29%, Firefox: 28%, Safari: 24%, and IE: 16%. Looking at my stats it doesn’t appear Google Chrome’s numbers were affected by pre-fetching. I guess Google’s promotion of their browser on their search and other pages is paying off. In addition to being a nice browser. Google Chrome lost me as a regular user when they got rid of side tabs and I couldn’t find an add-in I liked to replace them. I liked Chrome, but I really want the side tabs. This happened at the same time Firefox made their browser more stable and able to handle the 35 tabs I currently have open without crashing or sucking up all the memory. So it was back to Firefox, despite their annoying upgrades.
I’m still testing out different things on my Debian web server and will probably make a few more changes. My Mac Mini still has some disk performance issues so I’ll be looking at that and probably re-install everything like I did with my Windows 7 desktop. Then there’s Windows 8. It’s probably about time I install it to virtual machine and take a look. That should be enough to keep me busy.