I’m unable to boot an Amazon EC2 g2.2xlarge or g2.8xlarge instance using the Red Hat Linux 7 AMI. How can I fix this?

The Nouveau driver that is included in the Red Hat Linux 7 AMI cannot recognize a G2 instance type hardware configuration at boot.

During launch or reboot, the g2 instances might hang and display messages in the Amazon EC2 console similar to the following:

-----------------------------

[    352.217063] Modules linked in:[    352.231088] BUG: soft lockup - CPU#4 stuck for 23s! [migration/4:284]

[    352.231088] Modules linked in: adm1021 lm90 nouveau(+) intel_rapl crc32_pclmul ghash_clmulni_intel aesni_intel lrw gf128mul glue_helper ablk_helper mxm_wmi cryptd cirrus wmi syscopyarea sysfillrect sysimgblt video i2c_algo_bit ttm drm_kms_helper ppdev drm parport_pc parport i2c_piix4 pcspkr i2c_core ip_tables xfs libcrc32c ata_generic pata_acpi xen_netfront xen_blkfront ata_piix crct10dif_pclmul crct10dif_common libata crc32c_intel serio_raw floppy [    352.231088] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000

[  352.231088] Call Trace:

[  352.231088]    [] ? cpu_stop_should_run+0x50/0x50

[  352.231088]    [] cpu_stopper_thread+0x88/0x160

[  352.231088]    [] ? __schedule+0x2d8/0x900

[  352.231088]    [] smpboot_thread_fn+0xff/0x1a0

[  352.231088]    [] ? schedule+0x29/0x70

[    352.231088]    [] ? lg_double_unlock+0x90/0x90

[    352.231088]    [] kthread+0xcf/0xe0

[    352.231088]    [] ? kthread_create_on_node+0x140/0x140

[    352.231088]    [] ret_from_fork+0x58/0x90

To resolve the issue, disable the Nouveau driver at boot using the grub2 blacklist by following these instructions:

1.    Resize the instance, choosing any instance other than one in the g2 series.

2.    Edit /etc/default/grub and add the following values to the GRUB_CMDLINE_LINUX line:

rd.driver.blacklist=nouveau nouveau.modeset=0

3.    Rebuild the grub configuration:

grub2-mkconfig -o /boot/grub2/grub.cfg

4.    Stop the instance and resize it to your desired size same procedures as in step 1.

Note: A GPU instance must use the appropriate NVIDIA CUDA toolkit to enable increases in computing performance and take advantage of the graphics processing unit power.

Route 53, private hosted zones, VPN


Did this page help you? Yes | No

Back to the AWS Support Knowledge Center

Need help? Visit the AWS Support Center

Published: 2016-08-12