[GitHub] cloudstack pull request #1913: CLOUDSTACK-9752: [Vmware] Optimization of vol...

[nvazquez <gi...@git.apache.org>]

Github user nvazquez commented on a diff in the pull request:

    https://github.com/apache/cloudstack/pull/1913#discussion_r100543231
  
    --- Diff: plugins/hypervisors/vmware/src/com/cloud/storage/resource/VmwareStorageProcessor.java ---
    @@ -1577,11 +1577,15 @@ public Answer createVolume(CreateObjectCommand cmd) {
                     }
     
                     synchronized (this) {
    -                    // s_logger.info("Delete file if exists in datastore to clear the way for creating the volume. file: " + volumeDatastorePath);
    -                    VmwareStorageLayoutHelper.deleteVolumeVmdkFiles(dsMo, volumeUuid.toString(), dcMo);
    -
    -                    vmMo.createDisk(volumeDatastorePath, (int)(volume.getSize() / (1024L * 1024L)), morDatastore, vmMo.getScsiDeviceControllerKey());
    -                    vmMo.detachDisk(volumeDatastorePath, false);
    +                    try {
    +                        vmMo.createDisk(volumeDatastorePath, (int)(volume.getSize() / (1024L * 1024L)), morDatastore, vmMo.getScsiDeviceControllerKey());
    +                        vmMo.detachDisk(volumeDatastorePath, false);
    +                    }
    +                    catch (Exception e) {
    +                        s_logger.error("Deleting file " + volumeDatastorePath + " due to error: " + e.getMessage());
    +                        VmwareStorageLayoutHelper.deleteVolumeVmdkFiles(dsMo, volumeUuid.toString(), dcMo);
    --- End diff --
    
    Hi @syed @karuturi @SudharmaJain,
    What do you think about this approach? We will delete vmdk file if either createDisk or detachDisk fails, this way we make sure file doesn't exist in case that CloudStack retries operation. By the way, in VolumeOrchestrator lines 556-588 there's the retry logic, to retry only if failure contains "request template reload":
    
    ````
    for (int i = 0; i < 2; i++) {
                // retry one more time in case of template reload is required for Vmware case
                AsyncCallFuture<VolumeApiResult> future = null;
                boolean isNotCreatedFromTemplate = volume.getTemplateId() == null ? true : false;
                if (isNotCreatedFromTemplate) {
                    future = volService.createVolumeAsync(volume, store);
                } else {
                    TemplateInfo templ = tmplFactory.getTemplate(template.getId(), DataStoreRole.Image);
                    future = volService.createVolumeFromTemplateAsync(volume, store.getId(), templ);
                }
                try {
                    VolumeApiResult result = future.get();
                    if (result.isFailed()) {
                        if (result.getResult().contains("request template reload") && (i == 0)) {
                            s_logger.debug("Retry template re-deploy for vmware");
                            continue;
                        } else {
                            s_logger.debug("create volume failed: " + result.getResult());
                            throw new CloudRuntimeException("create volume failed:" + result.getResult());
                        }
                    }
    
                    return result.getVolume();
                } catch (InterruptedException e) {
                    s_logger.error("create volume failed", e);
                    throw new CloudRuntimeException("create volume failed", e);
                } catch (ExecutionException e) {
                    s_logger.error("create volume failed", e);
                    throw new CloudRuntimeException("create volume failed", e);
                }
            }
            throw new CloudRuntimeException("create volume failed even after template re-deploy");
    }
    ````
    
    To preserve this logic is that I've passed caught exception message on thrown exception in line 1587. Do you agree with this solution?


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---