May 10, 2008: Mail admin fix Qmailadmin was recompiled today to fix broken spell checking and a few other minor bugs. Mail administration login was temporarily disabled during the update, however email delivery was unaffected.

February 20, 2008: Email updates All email servers were updated yesterday to allow several new features including; domain specific SMTP IP addressing and DomainKeys digital signatures (email content and source authentication) for outgoing mail. Mail servers were down for only a few seconds and service was not affected by the upgrade.

February 13, 2008: Software Upgrades Most shared hosting servers were updated this afternoon to install Apache (1.3.41), add IonCube loader and Image Magick (PECL) support as well as tweak FastCGI options. This update required Apache and PHP be recompiled and caused 10-20 seconds of down time while PHP and Apache were restarted.

October 22, 2007: So Cal Fires While hosting services were not directly affected by the California fires, several employees were evacuated and doing business from motels while waiting for mandatory evacuations to end. Evacuations ended as of the first week of November and we appreciated your patience as new hosting orders were delayed during evacuation.

September 22, 2007: Webmail change RoundCube has replaced SQWebmail as our basic webmail program. If you use http://your.domain/webmail, you will now be directed to RoundCube. If, for some reason, you prefer SQWebmail, you can still reach it at http://your.domain/oldmail. Squirrelmail is still available at http://your.domain/mail.

September 15, 2007: Software upgrade Servers running Apache 1.3x have been updated to version 1.3.39 with simultaneous updates for eaccelerator, mod_ssl, mod_security and other security software. Some users may experienced up to a minute of PHP downtime as various PHP modules had to be recompiled before Apache was restarted.

August 5, 2007: Maintenance complete DB1 was rebooted this morning after the addition of a new 100GB primary hard disk partition. The reboot caused approximately 3 minutes of service downtime and caused no data loss.

July 31, 2007: Scheduled maintenance Servers DB1 and DG1 will be rebooted on Sunday August 5th in the early A.M. for primary drive partition resizing which will increase available disk space by over 100gb. The reboot itself will take only a minute or so, however startup disk checks are variable and may take up to 10 minutes to complete.

July 20, 2007: Network Outage A brief network outage occurred when an Equinix employee pinched our Mzima fiber when doing nearby work, cutting off primary SRO IPs. Mzima IPs are used only for shared hosting, so dedicated customers and shared customers with dedicated IPs were not affected. The feed is being rerouted to be "more tech monkey proof".

July 15, 2007: Software update PHP 5.2 is now installed on all shared servers. Customers have the option to use either 4.4.x or 5.2.x. Our implementation of PHP 5.2 requires different file permissions than 4.4, so you may need to contact support for conversion on an existing account. [More info]

July 7, 2007: Greylisting Although we've been experimenting for many months, this week we officially integrated Greylisting into our email system for spam prevention (see 'Mail Botnets' below). For more information about greylisting, please see this Wikipedia entry and this whitepaper. In a nutshell, greylisting delays the first email from any unknown IP address until the sender tries again. Valid SMTP servers always try again. Most bots will not. Bots which do try again will often be blacklisted before they get back. Thus, a very large quantity of bot generated spam is outright blocked. For more details about SRO Hosting spam prevention methods, please read this page.

June 22, 2007: DDos/routing issues At 8:10 this morning, we received notice from network monitors that some IPs were unreachable. We notified network engineers and though there were two more outages, the problem was resolved around noon. These outages appear to have been the result of a DDos attack against our network providers. The attacks were not directed against SRO and technicians were able to mitigate the issue. SRO servers continued operating normally throughout with no data loss. One positive side effect (in a quote from a network provider): "We have allocated immediate resources and directed engineering to upgrade our edge routers to add significantly more capacity and protection for our clients. These changes and upgrades will assure the network can handle over 10 times its current capacity."

June 1, 2007: Mail Botnets The 'bot problem' is getting worse. Zombie PCs connect to our servers several times a second to send spam. Anti-spam software stops most bots, however each connection still uses resources, so to reduce load we firewall bots so they can't connect. Easy, right? Unfortunately, millions of IPs in a firewall also creates a huge load, so we must block subnets. As a result, most residential IPs will eventually be blocked from port 25 on our servers. The 'right' solution would be for ISPs to (gasp!) shut down bots on their own networks, but that isn't going to happen, so the important news for SRO customers with a residential net connection is: stop using port 25 to send mail through SRO servers. Use 'SMTP submission' (port 587) or SMTPS (port 465) instead. Mail on these ports requires authentication and does not need to be firewalled.

May 14, 2007: Server maintenance We apologize that last night's maintenance took quite a bit longer than we would have liked, however it was still quite successful. In the space of a few hours, several servers had hard drives replaced and memory upgraded and while there were a few glitches, everything seems to be running smooth now.

May 7, 2007: Scheduled maintenance We will be performing multiple server maintenance and upgrade tasks during a scheduled maintenance window on Sunday, May 13th between 7 and 10pm PST. Due to the time it takes to mirror hard disks, this may result in site outages of up to two hours on customer servers undergoing drive upgrades. The SRO hosting website will remain online during these outages and any critical information will be posted here.

April 26, 2007: Minor bug fixes Nice, slow news month. We fixed a problem with mod security rules and also fixed that control panel slow login issue. We will be doing hardware upgrades in May which may result in up to an hour down time (but also increase performance and capacity) on several servers. Exact date/time will be posted when available.

March 12, 2007: DST Changes Sorry for the delay in updating for DST changes. We downloaded the new US zone info over the weekend, but failed to configure our servers to actually use the zone updates. If they throw more pointless DST changes at us in the future, we will automate the zone data update process to occur at the prescribed adjustment time.

February 23, 2007: PHP problems OK, so PHP 4.4.5 did cause problems. In fact, it rendered some scripts inoperable due to a session handling bug. We patched affected servers to 4.4.6RC2 and the bug is now gone (update: 4.4.6 stable now installed).

February 17, 2007: Software upgrade Servers running PHP 4.x have been upgraded to current version 4.4.5 . This is a minor update (addresses security issues and bugs) and should not cause compatibility problems. As always, let us know if you have any issues.

January 30, 2007: Firewall glitch Our apologies to those unable to use SSL mail early this evening. We re-configured most server firewalls over the last two days and neglected to open ports 465 and 995 on DB1 during the update, blocking SSL mail service.

January 9, 2007: Fuzzy Captcha! Supid headline, but a great addition nonetheless. All servers are now equipped for 'Fuzzy' OCR. This means most 'Image Spam' which can be read by a human can also be rated by our spam scanners. Combined with distributed spam recognition databases, per-user bayesian filtering, intelligent SPF, reverse dns analysis, smtp Dos mitigation, optional grey-listing and more, we are proud to have one of the most advanced spam management systems in the world.

December 23, 2006: Software Upgrades Sorry for the lack of 'good' news items recently, however over the last month we have in fact updated numerous software packages including Proftpd 1.3x (faster, security fixes), fastcgi, over a dozen new or updated Perl modules and Spamassassin 3.1.7 (more accurate, better per-user configuration). Most of these changes will be seamless, however a few of the Perl module updates might cause issues with (very) old scripts, so please let us know if you have any trouble.

December 11, 2006: Delayed mail delivery While working on DNS issues related to the previous announcement, we inadvertently created a bad DNS record resulting in customers on DB1 having trouble receiving mail today between 9:30am & 4:00pm. This has been fixed and the delayed messages should be arriving (rather quickly) as DNS changes propagate. No mail should be lost as a result of this incident, however we sincerely apologize for the inexcusably long time it took to detect and fix this problem. We will immediately begin work on a new monitoring system to detect mail delivery problems from outside our network as we currently do with http monitoring.

December 5, 2006: DNS update In response to outages caused by the recent attacks on everydns.net's servers (see below), we have been slowly migrating all SRO managed domains to use ns0.srohosting.com for primary DNS service instead of everydns.net. We will continue to use everydns.net's excellent services for geographic diversity.

December 1, 2006: DNS outage Our secondary DNS provider "everydns.net" was DDOS attacked today and their servers were occasionally slow to respond. Domains using ns0.srohosting.com for primary DNS should not have been affected, however domains using ONLY everydns might have appeared down at times. If your manage your own DNS, please be sure you have ns0.srohosting.com as primary DNS server. If not, please change this and let us know ASAP.

October 29, 2006: Short Outage Our primary ethernet switch for servers DB1 and DG1 went down this evening at around 6:30pm and was out for about 15 minutes before it was brought back up again. The DB1 server also lost time synch temporarily for some reason and was running an hour late (DST reverted) until about 7:15pm when we resynched it. This may affect some customer databases, so let us know if you need any help with time adjustments.

October 22, 2006: Datacenter outage Datacenter power circuits went down at 12:30 am this morning due to a botched UPS upgrade. All servers restarted normally, however there were two and a half hours downtime while engineers worked on the power. It is infuriating to have near 100% network and server up-time for years on end and see the datacenter ruin it due to pure incompetence. If this happens again, we'll be moving servers to a new datacenter very quickly.

October 15, 2006: Network maintenance Network engineers will be replacing a core router on Sunday, October 15th at 1:00 am. About 1/2 hour network downtime is expected.

August 26, 2006: Scheduled DC Outage As scheduled earlier, all servers were moved to a new power circuit on Saturday morning between 7:00 and 8:00am PST. This was done to increase resistance against future LA power issues (via datacenter UPS upgrade) The transfer resulted in an extended 1hr, 12min outage ending just after 8:00am with power cycling of ALL servers and disk checks. All servers returned to service without incident.

August 25, 2006: POP server outage POP email retrieval was down for about an hour this evening due to problems with an authentication module. The issue has been fixed for the moment, however we will be working on the POP server again later this evening and temporary outages are still possible. This issue only affected POP login. Incoming and outgoing mail has NOT been affected.