Table of Contents
Known Issues
Eucalyptus version 2.0.1
-
In certain cases, instances may remain indefinitely in "pending" state. You may find in cloud-error.log an error message related to java.lang.OutOfMemoryError: PermGen space. To resolve this issue, add the following to CLOUD_OPTS in eucalyptus.conf (this is the recommended size):
-XX:MaxPermSize=256mm - euca-get-console-output does not return the console output if the hypervisor used on the nc is Xen. To get the console output, log in on the nc and use the xm console command.
- If an instance is started without a public ip (i.e., using the --adressing private option), then an elastic ip is associated and subsequently disassociated with it, the instance will end up with a public ip.
- In MANAGED and MANAGED-NOVLAN modes, due to a limitation of iptables chain names, if very long security group and/or long login names are used, the rules associated with such security groups will not be effective, which tends to manifest as an inability to properly interact with instances (i.e., ssh into the instance will fail). For example,the login name graziano.obertelli@eucalyptus.com and the security group named default will trigger this issue.
Eucalyptus version 2.0.0
- Squeeze debian package installation/upgrade can exhibit odd behavior due to underlying dependencies changing often (For example, availability of the Xen hypervisor package by default). If upgrading, please see Upgrade Procedure for instructions on how to back up your existing data (users, images, keys, etc.) before attempting the upgrade.
- If the front end runs out of disk space, it can deadlock and hang. Make sure that you always have enough disk space for the front end logs, buckets and volumes.
- If the permissions on the buckets storage root directory for Walrus are too restrictive and "EUCA_USER" (typically eucalyptus) is unable to create files, Walrus may not respond to requests. Please make sure that the eucalyptus user can write to the buckets (and volumes) root directory.
- The file axis2c.log contains errors:
[error] rampart_handler_util.c(241) [rampart][rampart_handler_utils] 0 parameter is not set. [error] error.c(94) OXS ERROR [x509.c:284 in openssl_x509_get_subject_key_identifier] oxs defualt error , The extenension index of NID_subject_key_identifier is not validThese errors are benign and can be ignored.
- On some distros (CentOS), if you are using Eucalyptus with Xen, you might see an error like,
[EUCAERROR ] libvirt: cannot read directory /etc/xen: Permission denied (code=38)
This does not affect the node and can be ignored.
- If you see the following error in nc.log
[EUCAERROR ] libvirt: POST operation failed: xend_post: error from xen daemon: (xend.err 'Error creating domain: (4, \'Out of memory\', "xc_dom_boot_mem_init: can\'t allocate low memory for domain\n")') (code=11)
you are likely running a 64-bit image on a 32-bit installation. Please check your image.
- If nc.log indicates
failed to create swap.
you are likely out of disk space. Check that you have enough space under "INSTANCE_PATH" on the NC.
- Changing networking modes and most networking configuration options requires a CC (and sometimes a CLC) restart ($EUCALYPTUS/etc/init.d/eucalyptus-cc cleanrestart).
- EC2 command-line tools fail with Server: An error was discovered processing the <wsse:Security> header. (WSSecurityEngine: Invalid timestamp The security semantics of message have expired) Solution: Ensure that the clocks on the client and server machines are synchronized. This is not a Eucalyptus bug, but a consequence of the security policy enforced by the ec2 command-line tools.
- Difference from EC2: a Eucalyptus instance can only be a member of one security group.