Monthly Archives: February 2011

Engine Still Running Behind Schedule

After the changes we made yesterday to spread out the submissions over a pool of 10 separate IP addresses, the result has not been what we had hoped for and the submissions are still falling behind. We are therefore having to explore other options to try and speed things up.

New Servers Now In Testing

We’ve spent the weekend configuring our engine to run on multiple servers with multiple unique IP addresses in the hope that this will allow the submissions to run faster and start catching up with the backlog. This solution is based on the assumption that requests are timing out due to the increased number of concurrent requests made to each website on a single IP address.

We’re going to leave this new setup running over night and see how things have worked out in the morning. If the engine is still not catching up with the backlog, we’ll need to re-think again! But at least now we’re running over a pool of 10 different IP addresses which is better than 1 so it is still an improvement.

Again, thank you for your continued patience and please let me assure you again that all the jobs are still in the queue – no data has been lost!

Server Farm Now Begun

Following on from the previous post about the engine running behind schedule, we have now started work on our server farm. To begin with we are just running one additional server and wish to test this for a while to ensure everything works as it should before adding more. Assuming all goes well overnight, we’ll start adding more servers tomorrow. Once we have several in place they will just need a little tweaking to determine the best configuration for balancing the submissions amongst the servers but once done, the engines should then be able to catch up with the backlog.

[UPDATE 4.06pm GMT 11th Feb] It’s taking a little longer than we’d hoped to get the new servers up and running. Having two running has helped a little but it is not enough yet and the engine is still behind schedule. We’re continuing to work on this over the weekend and are hoping to have all issues fully resolved by Monday. We will of course continue to post updates here.

Thanks for your patience!

Submission Engine Running Slow

Over the last two days our submission engines have been running more slowly than usual which is causing it to get behind. This is very unusual as since the re-write they should be able to perform around 60 times faster than the previous engines! Investigation is underway as to the cause of the delay. In the meantime, please bear with us – all the jobs are still there! Note that secondary submissions have a lower priority than regular jobs so these are not being processed whilst the engine is in a backlog.

UPDATE [10th Feb 12.12pm GMT ] Upon further investigation of this on-going issue we are now suspecting that there could be two causes for the slow down. The first is a duplication bug which we have now eliminated but there may be a larger problem caused by the sheer amount of work being performed from a single server which has increased significantly over the last few days after the doubling of the Pro members!

We suspect that it is possible that we are making multiple concurrent submissions to any single site from the one server and although our server can make hundreds of submissions every minute, the remote sites that we submit to do not like that and are possible blocking (at least temporarily) our requests which would account for the incredibly slow response times from the sites. To give you an idea – a request that would normally take 5-30 seconds, is now taking several minutes in some cases!

If this is the case, it means we need to get our multiple server farm up and running as quickly as possible! We could restrict submissions in some way as a temporary measure but we don’t want to reduce the level of service offered and it will be much easier to diagnose the problem if the engine continues to be subjected to the same degree of workload.

Therefore, operation will continue as normal whilst we continue to work on the issue. Currently, the engine is running around 19 hours behind schedule for regular submissions and secondary submissions are not being processed at all whilst the engine is in backlog. Please bear with us whilst we continue to improve our engine!

SPAM Filter taking shape

A couple of weeks ago we wrote a post about the spam issue. As a recap: with the large number of members now using IMAutomator, it is inevitable that some people are posting materials (bookmarks mainly) that are causing our accounts to get deleted and in some severe cases, our server is being blocked entirely. Obviously this hurts everybody who uses the IMAutomator system.

Upon reviewing the policies of the bookmarking sites we submit to, we have come up with a set of acceptable use guidelines – or rather, areas to be avoided. These are adult, gambling, drugs and anything illegal.

To get rid of the spam, we need a two-pronged attack. Firstly the submissions that are already sitting in our job queue waiting to be submitted must be cleaned, and secondly we need to prevent such submissions being made in the first place. Our first priority is in cleaning up the queue to ensure that no more damage is done.

We’ve begun to develop our internal spam filter and at the moment it is a little crude but over time it will learn what is spam and what isn’t. We’ve introduced a new submission error message of ‘Deleted for SPAM content’ though this doesn’t indicate what triggered the error. As we refine the spam filter we will be able to report on the nature of the content such as whether it was adult etc.

The next stage will be to actually prevent the submissions from being made at the point of entry on the submission form but we will not implement this until the spam filter has learned what is spam to a very high degree of accuracy as obviously a false identification would be very frustrating for members.

Note that the amount of spam is actually higher than we had realised and so this has now become a top priority and we are going to hold off adding more Bookmarking sites until this is fully resolved. We don’t want to go to the effort of adding a PR6 site only to have our server blocked a week later due to spam!

Cleanup Run on Problem Sites

In the last post I explained that a problem with a few sites had caused several failures. The sites in question were as follows: SOCIALARTICLE, PLANETBLOG, FLASHAHEAD, HOSTINGTAGS, BUZZBOOKMARKS, ROLLINGTAGS.

We have now rectified the problems so the sites are working as normal now and although we cannot fix the submissions that have already failed, we’ve been able to offer a solution of sorts for Pro members using the Secondary Submissions feature.

Secondary submissions allows jobs to be resubmitted to any new sites that were added since the original submission was made. What we have done is to remove the submission record for the failed submissions to those sites for Pro members so that when the Secondary Submissions report is run, jobs will be re-submitted to those sites so they will get a second chance.

As this feature is only available to Pro members we have not modified the data for Free members and can only apologise for the initial mistake that was made with the sites.

Problem with a few Bookmark sites

On Monday when we launched Release 2, we added a lot of new Bookmark sites to the system. A mistake was made with a few of them which caused them to get added to the live system before they were completely ready which has caused errors on most of the submissions.

Although we cannot fix the submissions that have already been made, we are working quickly to fix the broken sites and for Pro members we can remove the failed submissions so that they can be re-scheduled using the Secondary Submissions feature.

We hope to have this sorted by Monday and will post again when they have been fixed properly and a cleanup run. Apologies for the inconvenience!

Automatic Article Bookmarking – An Integration of Tools

Now that the article submitter has been in place for a little while now we’ve been taking a closer look at how these directories work. Some of them approve articles quite quickly while others seem to take weeks which can be a nuisance.

However, it should be possible for us to construct the actual URL of the article created at each site and then run a report to check which ones are live. This could be integrated into the system and would then allow those Article URL’s to be bookmarked if you so desired.

This would provide a second layer of backlinks to your original content. This is not as powerful as the direct backlinks but could provide if used with the more prominent article directories in particular – such as Article Dashboard.

Most Pro members don’t use up all 25 bookmarks per day so this could provide a nice way of integrating the tools and get even better value from your membership.

NO MORE SHOUTING!

A minor tweak has been made to the validation routines today to check for submissions where titles, descriptions etc have been made in ALL CAPITALIZED CHARACTERS. This is commonly known as ‘SHOUTING’ and is one of those ‘netiquette’ things that can cause something to be regarded as spam.

You can still enter the odd capitalized word but if there is too much in there (around 50% or more) then you’ll be given an error message.

Release 2 Sold Out in 4.5 Hours! What’s Next?

Yesterday’s response to the launch of Release 2 was amazing and it sold out in just 4.5 hours which is somewhat quicker than Release 1 which tool a whole month to fill up! I know this has led to some disappointments as not everybody who wanted a place was able to get one but every now and then an existing member cancels which opens up a place for somebody else.

I’d like to welcome aboard all the new Pro members and point you towards our Training page to make sure you get the best out of all the tools you now have available!

So with Release 2.0 done, what’s next? Theres a LOT of work to do, and here is a very brief highlight of some of the major developments on the way:

  • Combating spam – we need to start filtering out materials that are causing our accounts to get deleted and our server to be blocked. This means no adult material, viagra, gambling, warez etc – it has to go and this is our first priority!
  • Multiple server farm – we’re expanding in a big way so instead of having just one server running our submission engine, we’ll have a whole bunch of them! This also allows us to do things such as have priority servers for Pro members.
  • Article submission improvements – we’re going to implement a system that allows the size of the article to control the number of sites that are submitted to. So for example, a 300 word article might not be able to be submitted to all the sites as some have a 400 or 500 word minimum for example, but it can still benefit from some. This is preferable to just setting a blanket minimum for all articles.
  • Spinning / Rotation – this was supposed to go into Release 2 but we pushed it back to Release 3 because it applies particularly well to articles as well as bookmarks. This feature will allow you to generate multiple unique articles, titles, descriptions etc for your submissions. We’ll be liasing with members to see how people want this implemented as there are many options.
  • Lots more sites! Before we start work on the 2 new submission tools for Release 3 we want to get in a lot more sites for our existing tools. This will be a steady, ongoing process!

Stay connected with us in your favorite flavor!