Over at the O’Reily Port 25 blog Todd Ogasawara asks the question in the title of this post. The larger context is that the Fedora (and Red Hat) servers were compromised recently and if you’ve installed either the software or any patches you need to at least be considering the implications of these security breaches. But that’s not what I want to comment on.
Does Mr Ogasawara really live in such a cocoon as to not know that a lot of production servers probably run Fedora and other free linux distributions???? And there is one reason for this, the software is FREE. Free means no one has to sign a purchase order for the software, free means a bright person can build a cool new website without outlaying any money up front (assuming the have 1 piece of hardware). Also the people using Fedora as a production probably aren’t the target market for the ’support’ or ‘we need someone to sue if anything goes wrong’ camps.
Aside from free you know another reason people use Fedora and other free distros for production? Apps and tools. Personally I run Solaris (also free for download) but honestly every time I have to figure out why tool X won’t compile on it I’m tempted to migrate to a linux distro where everything works (mostly) out of the box.
Finally I will add one more thing. Having used Red Hat Enterprise I really don’t see what value it adds for the majority of users (i.e. companies) that purchase it. I mean besides a number to call to help you support it. But I think you’re money (in a lot of cases) would be better spent hiring bright people who can get the free distro working with whatever problem you are having.
This past Saturday I retired my old Soekris Net4501 running m0n0wall in favor of a home built PC running pfsense.
I was motivated not because of m0n0wall had failed me in any way. In fact it has been great. Instead I need 2 things that made me replace m0n0wall and the Soekris net4501, multiple WAN fail over and 4 ethernet ports.
I had some trouble finding the right combination of hardware from my spare parts to get the machine up, but once it was up and installed things went quickly with only a minor hitch. That minor hitch was with my initial load balancing configuration AIM did not work at all. To fix that I just enabled ‘Sticky Connections’ and was all set.
It has now been running successfully for over a day and in reviewing the logs my DSL link has dropped a few times without any noticeable loss in connectivity.
Covad is making another call to look at it tomorrow after noon (12-4). Looking at my graph (like the one I post previously) the DSL line was basically bouncing up and down from around 8am until 4pm since then though it has been mostly stable. AI sure hope Covad can find the problem tomorrow. I would also love, love, love for the problem to be attributable to Verizon and they have to fix it.
But in good news I now have a back-up link installed so I can now tolerate downtime on the DSL link.
Oh the wireless install wasn’t as easy as I was hopping for. The physical installation went just fine it was getting a computer online that proved more difficult. The technician who did the install was great but not very tech savvy. He mentioned the link was a PPPoE connection. So we tried to set up a PPPoE connection on my Mac. That didn’t work. We were able to get his Vista laptop working just fine so we knew the link was good. He called back to the help desk and with their help got me up. As it turns out it isn’t really a PPPoE link as far as I’m concerned. It is just straight ethernet.
My DSL woes continue. I received a new modem from Speakeasy Tuesday but that hasn’t helped the problem. Here is a somewhat typical outage period:
5 seconds DOWN
16 seconds UP
35 seconds DOWN
128 seconds UP
5 seconds DOWN
64 seconds UP
62 seconds DOWN
3 seconds UP
39 seconds DOWN
18 seconds UP
32 seconds DOWN
73 seconds UP
32 seconds DOWN
5 seconds UP
46 seconds DOWN
19 seconds UP
37 seconds DOWN
68 seconds UP
4 seconds DOWN
9 seconds UP
30 seconds DOWN
20 seconds UP
That makes it particularly hard to do just about anything online. Covad is spoosed to send a tech out (which might be billable if they don’t find anything wrong) but I haven’t heard back when that will be.
August 11th, 2008DSL woes
For about the past week (since the thunderstorms rolled through Hampden and burned down a church) my DSL line has been less than stable. A week ago Saturday the line was completely down. Sunday it came back and until Tuesday it looked like everything was improving. Then I started noticing that the line was dropping and while monitoring ping packets and instead of getting better over time the overall trend was getting worse.
On Friday I had decided to contact Hampden Wireless to set up service for a back-up internet connection for times when my primary Speakeasy DSL line goes done in the future. Sadly I’m in a dead spot for their service. Although after their email address on the website bounced and seeing the inside of their store front (Little Shop of Hardware) I’m not terribly upset.
By Sunday things deteriorated to the point where, when Speakeasy followed up on my open ticket I explained what was happening and we determined that a new modem might be a good idea to narrow the problem. After speaking with them it seemed like the DSL was getting less stable. Around 3:50pm I decided to eliminate my internal household wiring and plugged a phone line directly into Verizon’s ‘demark’ box outside. This eliminated my internal wiring as the problem but didn’t solve the problem. In fact while I don’t think it made anything worse it certainly didn’t help and correlated to a noticeable period of instability of on the DSL line. Later in the evening it stabilized for a while, long enough for me to install hobbit (Big Brother’s replacement/current incarnation) to better monitor the line. And here is the result from monitoring the line today:
Each red bar indicates an outage…..so while there were periods of stability.
After yesterday’s frustrations I resolved to find another back-up connection (and NOT Comcast). Luckily I think I found one in Believe Wireless. I contacted them during the day and they will be out Friday to install an antenna for service. For the time being I’m just signed up for their most basic service to use as a back-up link.
Oh and speakeasy has sent me a new modem to see if that solves the DSL issues. If not it will involve a dispatch from Covad to hopefully trace the problem.
Oh and despite my DSL woes I’m not angry with Speakeasy.
Maybe is is just the spammers targeting me but so far in August I’ve received just 1 real spam message to my email account. My server has even been rejecting less over the last couple months as you can see in this image:
I’d like to hope this will continue but probably won’t. Oh well I will enjoy it for as long as it lasts.
Early Saturday morning we had a thunderstorm roll through Hampden. A lot of lightning strikes very close by. (One actually struck and set a church on fire couple blocks away). We had DSL connectivity initially Saurday morning but lost it at about 9:30ish. Shortly after 10am I determined that the DSL link was downa nd it wasn’t just a networking issue. So I called my DSL provider, Speakeasy.
Let me take a moment here to say that calling thier support number is not a pleasure (because it means my DSL is out) but very pleasant. The support staff is knowledgeable, pleasant and genuinely want to help you not just get you of the phone. Sure they make you reboot the modem and plug and unplug things but they need this to run loop tests on the line. So it isn’t like they are just trying to make you feel like they are doing something. Oh and here’s one of the things that make it pleasant. They take ownership of a problem and actually call you back is they say they will. Now enough of my praise for Speakeasy support. Back to my problems.
George who answered my call had me do the normally reboot, unplug modem, ask if anything had changed in my environment. etc. He determined it was nothing on my end and opened a ticket with Covad to check the card in the DSLAM in the CO and see if it was the issue. He said they should address that ticket in about 4 hours. That was fine with me. I had plans for the afternoon so could wait a few hours. When 6pm rolled around and I hadn’t heard back from them I decided to leave the birthday party I was at in Roland Park and head home to see what was up. Once home I called to check the status of my ticket. I was informed that Covad wanted to do check in person and that this could be billable if they found nothing wrong with the DSL line to my house. Not ideal but I was confident enough at this point that it wasn’t my problem so I wouldn’t be billed. I requested 1st thing Monday but they couldn’t accommodate that so it was scheduled for first thing Tuesday. I went to bed with the DSL line still out.
Sunday we woke up to no internet again. Sunday afternoon I decided to do the last bit of testing I could and found a phone line long enough I could get from the telco box directly inside. I then relocated the DSL modem to connect to the new phone line. When I connected it there was no DSL connectivity. However later around 2pm I noticed the DSL light on the modem lit. So I plugged it back into the firewall and ta-da the internets were back! After we got our fix of internets I re-wired everything back to the way it was prior to Saturday morning. Everything is working again.
I can’t say for sure but I suspect Verizon is to blame for this outage. The outage occurred after a nasty thunderstorm and was later restored without Covad doing anything. I can totally see Verizon ‘re-wiring’ something to re-route around damage and interfering with my ‘naked’ DSL line.
Now I need to call Speakeasy, inform them my lines is back p and cancel the Covad service call.
I had a few errant imapds processes out sucking up CPU after playing with some new certificates last night. The result when I logged on to my server for another reason was this:
pjohnson@skyline:~$ w
4:24pm up 138 day(s), 20:18, 1 user, load average: 49.06, 48.86, 48.68
a quick ps and kill disposed of the imapds processes but it also cause a surprising spike in the load:
pjohnson@skyline:~$ w
4:26pm up 138 day(s), 20:20, 1 user, load average: 1146.27, 281.72, 127.09
That’s what I get to playing with things before I went to bed last night.
I’ve had my Drobo for just over a week now and have some early impressions I thought I would share:
slow
I noticed the drobo was not as quick in receiving data when I was first migrating my data over to it. I was hoping that was just because I was copying from 1 USB to the USB connected drobo. Sadly the slowness while better is still there and I have no other USB drives connected to the machine now. A lot of the reviews mentioned this but I was hoping that their perception of slow wasn’t as forgiving as mine can be.
resume after idle
The Drobo after a period f inactivity will idle (spin down?) the drives to save wear and tear as well as energy. I’ve been on a energy conservation kick lately so I can’t complain about this aspect. I will however note that this can be annoying. When I cd into the mount point on I have the drobo on and try to list files there is a 5-10 second delay before I get a response, while the disks spin back up.
concurrent access
I have to advise against using the Drobo for concurrent access by more than one task. In an effort to find and delete duplicate MP3s I have (some are exact duplicates, others same song at different bit rates) I’ve been running an perl script to gather ID3 tag information as well as using my Mac Mini to copy files and re-tag files and manually delete duplicates. When I’m copying and running the perl script I notice that both take longer. I looked at the output from top and neither smbd nor perl was consuming more that 20% of the CPU resources but the load was still high so my theory is that IO was what was driving up the load.
I’m hoping that my efforts to delete duplicate files pays off with some significant disk space saving otherwise I will be needing to expand the array sooner rather than later. I’m already at 76% capacity for my current array.
Despite the limitations above I like my drobo and still feel for my modest needs I made a good choice in disk storage technology that will allow me to easily grow as my disk space needs grow. I can’t recommend using it as online storage for anything either disk intensive or something you need to copy off quickly. I will say though it makes a great back-up device and storage repository if you can accept the slow speeds of data access.
I actually expect stuff like the following from the Baltimore Sun (considering they have like 6 people left working at actual reporting anymore) not from the Baltimore Business Journal.
Here’s the headline:Gen Y precedes Gen X in technology alphabet
According to a new report, Generation Y is a small but influential group when it comes to technology adoption, while Gen Xers choose to adopt a technology when it fits in with a personal need or desires
I might buy that but then I read this:
While Gen Y is a small generation of 18- to 28-year olds, consisting of only 38 million adults
and later this:
In contrast, Gen X, which is composed of 29- to 42-year olds — about 63 million adults
First of all they are comparing a generation that spans 10 years to one that spans 13. That alone would should have caught the reporter’s attention. That of course ignores the fact that a generation is usually minimally a 20 year span.
And in re-reading the article a 2nd time it reads more like a press release from the survey’s source: Forrester Research. What happened to reporting? You know like questioning the methodology or the conclusions drawn from the data?
Read it for yourself and see what you think.
