Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


Digitalocean snapshot, etc. experience
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

Digitalocean snapshot, etc. experience

Hi All:

I have been looking at digitalocean for a while and mostly I like it. But I have had some problems. I am wondering if anyone else has had similar problems.

The last time I took a snapshot of a 20G SSD droplet (VPS). It took over 2 hours to complete.
The exact time is 2 hours 36 minutes 1 second execution time. This looks to me way out of the ordinary. Something must be wrong. I had about 10G data on it. It just does not take that much time to create an image of 10G, zip it, write it to some file. I had taken snapshots before for similar sized droplets. It usually took just over 10 minutes. And the response from the support is not very helpful.

After 30 minutes of running the snapshot, I thought that there must be something wrong. I opened a ticket. The reponse came back to me fairly quickly and just told me to be patient.

Then after 60 minutes, I opened another ticket. And this time it was just ignored.

In the end, after the snapshot was completed (over 2 and half hours), someone at the support department emailed me and told me that perhaps I had too many files (inodes). But I had at most about 10,000 files. And I had taken snapshots of the same vps numerous time (10-20 times) before. And it always took about 12-15 minutes.

My feeling is that the support people over there just do not use their heads to think about a problem and investigate it if needed. All the responses I have had seem to be some automatic formalities.

For example, once I clicked to destroy a droplet and the process just hanged there. So I opened a ticket, "be patient" was the response. And telling me perhaps the "scrub data" option made it slow as it needed to write all these zeros to a disk. But if anyone has any experience with computers, he/she must know that it does not take long to write 20G of zeros to any hard disk (let alone SSD). So tickets after tickets of my telling them something was wrong that I had done it many times, should be almost instant, finally someone decided to take a look and found the process hung. So he/she had to intervene to kill the process and I had to re-destroy it.

Also lately, creating a droplet is taking a bit longer than before. The promise is about 1 minute. Now it takes about 5+ minutes to create a droplet.

I am just wondering if someone else had similar experience.

cheers,

ftc

Comments

  • well my droplets are created the fastest when i choose new york nyc3
    maybe that helps

  • LeeLee Veteran

    @freetochoose

    They use a queuing system, It gets bogged down at times, I see that too. Sometimes an hour to do a snapshot, other times 10 minutes.

  • We have few servers with them, snapshot takes time now and what i heard is they are using some service like Amazon S3 or other to store the snapshot. This helps them in region to region transfer, this is what i heard not sure about it. Because of this transferring of data it takes time.

    Concerned with the support, nowadays they are slow, not responding properly unless you put pressure, simply redirect you elsewhere, many times their site webpages show custom 404 error page informing that the " resource you are looking is not found" after refreshing it works.

    They may be overloaded owing to their popularity.

  • thanks all.
    but taking 2 hours 36 minutes 1 second execution time for a snapshot of a 20G SSD with about 10G data is ridiculous. and the automatic formal response is also bad because i was expecting a thinking human being doing the supporting.

  • Customers must understand how backups will be done. This decision must be made on the storage technology which are used to hold your data. Local and shared storage is important. The filesystem or storage types are also important. Filesystems like ZFS can create and hold snapshots easily on the same cluster. They can also use particular block copy to another cluster node.

    If your provider offer shared storage, they can create snapshots directly on the SAN over the hypervisor. This takes only seconds and will be created directly on the storage. Enterprise hypervisors like vmware or xenserver using this to create completely snapshot histories. This snapshot can be later transferred to a additional SAN cluster for archiving without any interruptions to the customer or hold on the primary storage. This snapshot is available in the complete zone which using the same shared storage, no problem to deploy this snapshot on other nodes within seconds.

    Providers with local storage must be done a few extra steps. They must transfer the snapshots between they nodes and backup servers because the nodes do not share the same storage.

    Easy example, CloudStack can create snapshots called VM snapshots on the primary storage within seconds because they don't need to copy the snapshot from the local storage to any other system. You can also do a snapshot do a second storage and also have the same issue to copy the image to another SAN cluster. Local storage backups can be speed up with technologies like template linking or transferring only different data from previous data based on raw blocks.

Sign In or Register to comment.