DreamObjects Review: My Bad Experience


I’ve just about had it with DreamObjects. I only had about 250GB of photos/thumbnails on the “West” server. I was told earlier in the year that it would need to be migrated to the new “East” server by October 1st, 2018. I heard there were bugs and glitches with the migration early in the Spring of 2018, so I delayed my migration until July.

It took nearly 4 months to migrate everything, mostly due to bugs/glitches. The migration tool in the panel didn’t work at all and was completely broken/inaccurate, so Dreamhost Support had to manually migrate it. Because I could not migrate it before October 1st, Dreamhost said I would still have access to the “West” server even after the deadline, and there would be no disruptions.

On October 9th, without warning, Dreamhost turned off “WRITE” access to the “West” server. We run two large marketplaces with thousands of merchants. This means that none of our sellers could upload new products with images, since uploading images to DreamObjects was not possible with “WRITE” access turned off.

Two days later (Oct 11th), I was told the migration of our DreamObjects buckets were complete, so I updated my DNS settings and our API to change the “West” server to “East”. After wasting time with DNS propagation with no images showing up on our websites, I found out that Dreamhost never migrated our “public-read” permission flag on our images, so all of our images had the “private” flag on them and could not be loaded up in browsers.

Another six days later (Oct 17), Dreamhost finished updating the permission flag so that all of our images could again load up for site visitors. However, sellers still have problems uploading photos to the new “East” server. Whenever they try uploading photos, the page freezes up for them and they have to refresh the page to re-attempt uploading images. Our logs show numerous “408 Request Time-out” errors and “500 Internal Server Error” errors. This has caused our image uploading scripts to timeout and our web server processes to be frozen until timeout as well (resulting in our whole website freezing up for visitors). Dreamhost still has no solution for catching exceptions for these errors to prevent the scripts from timing out because of DreamObjects server problems. Dreamhost had apologized to us and have credited us all of the money we paid for DreamObjects due to the numerous problems we’ve had with it for months. However, I would have preferred a working, efficient server.

We’ve kept a note on our marketplace for our thousands of sellers when they list items, that our hosting service is having problems with our photo uploading server. This note, and the errors and the site problems, have been ongoing without solution since October 17th until November 11th (today), and still there is no solution or fix from Dreamhost, even though we’ve received many messages of “investigation” and “escalation to higher priority”. Our sellers have been writing us almost daily reporting problems and asking when image uploading will be fixed.

Starting today and yesterday, many images from DreamObjects are not even loading up anymore. So not only is there problem with “WRITE / PUT” for sellers trying to upload new images, there is now “READ” problems that prevent sometimes half the images on our page from even loading on our marketplaces - again slowing the entire website down and requiring visitors/buyers to refresh the page multiple times for these images to load. The problems have only gotten worse. Not only are sellers complaining to us about not being able to upload new product images, but now buyers and site visitors are complaining about images not loading.

I’ve warned Dreamhost that I will be switching to a new hosting service if the problems are not fixed quickly. It’s nothing personal, I’ve always liked Dreamhost, but this DreamObjects problem is by far the worst problem I’ve had with Dreamhost in over a decade. Sellers on our site are very angry with the problems and I simply need this fixed immediately, even if it requires moving to a different hosting service.

My first pick to look into is “Spaces” by DigitalOcean: https://www.digitalocean.com/products/spaces/ . They are charging $5 for 250GB and 1TB download, and $0.02/GB space and $0.01/GB download afterwards. Does anyone have experience with them? Is this a good, reliable service or not?

Personally, I think DreamObjects should still be in Beta - never should have been in production with this many bugs and problems. Never should there have been an Oct. 1st deadline to migrate all buckets from “West” to “East” without Beta testing the “East” server and making sure everything was set to go - and making sure the migration tool would go smoothly.


23 days later (Dec 4th), still no solution or progress by Dreamhost support…

Hundreds of pissed sellers in my multiple marketplace cannot upload new products for sale during the Holiday season (cannot upload images). They can upload 1-2 images before the photo uploading script freezes because of Dreamhost’s object server constantly hanging/freezing for no apparent reason. Because of DreamObjects hanging, it causes the rest of my PHP processes on my DreamCompute instance to hang and timeout after 5 minutes (resulting in 500 Internal Server Errors for normal site visitors and Googlebot).

I’ve set up an empty test bucket with 1 image in it. Using the same simple tutorial examples (https://help.dreamhost.com/hc/en-us/articles/115002875812-Sample-AWS-PHP-v3-SDK-code), DreamObjects continues to hang/freeze on a simple “List Buckets” call. It works 1-2 times, and then you get the hang… Not sure what the deal is with Dreamhost. Maybe someone else has had this exact same problem and can offer a solution?


I keep meaning to reply, and then something hangs and I spin my wheels on that.
I don’t have your problem. I get listings, uploads are fine downloads work fine, all within an acceptably short delay. So I’m no help figuring out your issue. I do almost all of my uploads and lists in perl scripts. Downloads are mostly direct to browser https: I do download a few through the perl LWP::simple module and don’t seem to have any problems. Because I had problems with their migration tool, a large number of my files that were transferred by DH support. Those files were set to private, and I couldn’t get the perl Amazon::S3 to change their acl, so I used the command line “s3cmd setacl…” in a perl system statement. Or used a file browser for some. That was a disappointing time waste. Also on Dec. 4, I got an email saying they were done: “Your S3 migration from objects-us-west-1.dream.io did not complete successfully.”. That was a day or so after west became inaccessible. Actually, I think I did get everything transferred.

However, Just now, and one other time in the last couple days the cloud server has disappeared for several minutes at a time. Uploads stop and downloads timeout. It comes back in fits and starts; if there is active uploading at the time a few files get missed and I need to do some cleanup.
The off line event is not over. I had some connectivity for a few minutes but not now.


This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.