- Refactor cloud-early-config and make appliance specific scripts
- Make patching work without requiring restart of appliance and remove
postinit script
- Migrate to systemd, speedup booting/loading
- Takes about 5-15s to boot on KVM, and 10-30seconds for VMware and XenServer
- Appliance boots and works on KVM, VMware, XenServer and HyperV
- Update Debian9 ISO url with sha512 checksum
- Speedup console proxy service launch
- Enable additional kernel modules
- Remove unknown ssh key
- Update vhd-util URL as previous URL was down
- Enable sshd by default
- Use hostnamectl to add hostname
- Disable services by default
- Use existing log4j xml, patching not necessary by cloud-early-config
- Several minor fixes and file refactorings, removed dead code/files
- Removes inserv
- Fix dnsmasq config syntax
- Fix haproxy config syntax
- Fix smoke tests and improve performance
- Fix apache pid file path in cloud.monitoring per the new template
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
- Load the nf_conntrack_ipv6 module for IPv6 connection tracking on SSVM
- Move systemd services to /etc and enable services after they have been
installed
- Disable most services by default and enable in cloud-early-config
- Start services after enabling them using systemd
- In addition remove /etc/init.d/cloud as this is no longer needed and
done by systemd
- Accept DOS/MBR as file format for ISO images as well
Under Debian 7 the 'file' command would return:
debian-9.1.0-amd64-netinst.iso: ISO 9660 CD-ROM filesystem data UDF filesystem data
Under Debian 9 however it will return
debian-9.1.0-amd64-netinst.iso: DOS/MBR boot sector
This would make the HTTPTemplateDownloader in the Secondary Storage VM refuse the ISO as
a valid template because it's not a correct format.
Changes this behavior so that it accepts both.
This allows us to use Debian 9 as a System VM template.
Not sure though if enabling them is enough for systemd to still start them
on first boot
Signed-off-by: Wido den Hollander <wido@widodh.nl>
SystemVM changes to work on Debian 9
- Migrate away from chkconfig to systemctl
- Remove xenstore-utils override deb pkg
- Fix runlevel in sysv scripts for systemd
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
We found bug in ACL rules for PrivateGateway for VPC
At a glance - rules not applied - switching Allow All or Deny All (default ACL) - showed as completed - but rules missed.
Result - traffic via PrivateGateway blocked by next DROP rule in next chains
How to reproduce:
Enable PrivateGateway for Cloudstack
Create VPC
Provision new PrivateGateway inside VPC with some VLAN
Change ACL (optional step to show that problem not in initial configuration but in config itself)
Expected:
ACL rules applied (inserted) into correspondig ACL_INBOUND/OUTBOUND chanins for PrivateGateway interface (ethX) based on ACL which user choose
Current:
No rules inserted. ACL_INBOUND/OUTBOUND_ethX - empty. Traffic blocked by next DROP rule in FORWARD chain
Affect - all our corporate customers blocked with access to their own nets via PG and vice-versa.
Root cause:
Issue happened because of CsNetFilter.py logic for inserting rules for ACL_INBOUND/OUTBOUND chains.
We choose rule numebr to isnert right before last DROP rule - but forget about fact - that if chain empty - we also return 0 as insert position. Which not true for iptables - numeration started from 0.
So we need very small patch to handle this special case - if number of rules inside chain equal to zero - return 1, else - return count of rules inside chain.
It's found only one - just because be default for PrivateGateway - we didn't insert any "service rules" (if SourceNat for PrivateGteway not ticked) - and we have by default empty ACL_INBOUND/OUTBOUND chains. Because same insert happened for all VPC networks (but when we call this insert - we already have at least 1 rule inside chains - and we successfully can process)
When auth strictness is set to true, terminate SSH handshake for clients
that do not present valid certificates.
This uses the `setNeedClientAuth`, where if the option is set and the
client chooses not to provide authentication information about itself,
the negotiations will stop and the engine will begin its closure
procedure:
https://docs.oracle.com/javase/7/docs/api/javax/net/ssl/SSLEngine.html#setNeedClientAuth(boolean)
During systemvm reboot, the conf folder is removed and certificate
re-setup is not done. This may cause the agent to not connect, this
fixes the case by backing up and restoring keystore and other config
files when re-patching is done after rebooting of a systemvm (cpvm, ssvm).
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
This introduces a new certificate authority framework that allows
pluggable CA provider implementations to handle certificate operations
around issuance, revocation and propagation. The framework injects
itself to `NioServer` to handle agent connections securely. The
framework adds assumptions in `NioClient` that a keystore if available
with known name `cloud.jks` will be used for SSL negotiations and
handshake.
This includes a default 'root' CA provider plugin which creates its own
self-signed root certificate authority on first run and uses it for
issuance and provisioning of certificate to CloudStack agents such as
the KVM, CPVM and SSVM agents and also for the management server for
peer clustering.
Additional changes and notes:
- Comma separate list of management server IPs can be set to the 'host'
global setting. Newly provisioned agents (KVM/CPVM/SSVM etc) will get
radomized comma separated list to which they will attempt connection
or reconnection in provided order. This removes need of a TCP LB on
port 8250 (default) of the management server(s).
- All fresh deployment will enforce two-way SSL authentication where
connecting agents will be required to present certificates issued
by the 'root' CA plugin.
- Existing environment on upgrade will continue to use one-way SSL
authentication and connecting agents will not be required to present
certificates.
- A script `keystore-setup` is responsible for initial keystore setup
and CSR generation on the agent/hosts.
- A script `keystore-cert-import` is responsible for import provided
certificate payload to the java keystore file.
- Agent security (keystore, certificates etc) are setup initially using
SSH, and later provisioning is handled via an existing agent connection
using command-answers. The supported clients and agents are limited to
CPVM, SSVM, and KVM agents, and clustered management server (peering).
- Certificate revocation does not revoke an existing agent-mgmt server
connection, however rejects a revoked certificate used during SSL
handshake.
- Older `cloudstackmanagement.keystore` is deprecated and will no longer
be used by mgmt server(s) for SSL negotiations and handshake. New
keystores will be named `cloud.jks`, any additional SSL certificates
should not be imported in it for use with tomcat etc. The `cloud.jks`
keystore is stricly used for agent-server communications.
- Management server keystore are validated and renewed on start up only,
the validity of them are same as the CA certificates.
New APIs:
- listCaProviders: lists all available CA provider plugins
- listCaCertificate: lists the CA certificate(s)
- issueCertificate: issues X509 client certificate with/without a CSR
- provisionCertificate: provisions certificate to a host
- revokeCertificate: revokes a client certificate using its serial
Global settings for the CA framework:
- ca.framework.provider.plugin: The configured CA provider plugin
- ca.framework.cert.keysize: The key size for certificate generation
- ca.framework.cert.signature.algorithm: The certificate signature algorithm
- ca.framework.cert.validity.period: Certificate validity in days
- ca.framework.cert.automatic.renewal: Certificate auto-renewal setting
- ca.framework.background.task.delay: CA background task delay/interval
- ca.framework.cert.expiry.alert.period: Days to check and alert expiring certificates
Global settings for the default 'root' CA provider:
- ca.plugin.root.private.key: (hidden/encrypted) CA private key
- ca.plugin.root.public.key: (hidden/encrypted) CA public key
- ca.plugin.root.ca.certificate: (hidden/encrypted) CA certificate
- ca.plugin.root.issuer.dn: The CA issue distinguished name
- ca.plugin.root.auth.strictness: Are clients required to present certificates
- ca.plugin.root.allow.expired.cert: Are clients with expired certificates allowed
UI changes:
- Button to download/save the CA certificates.
Misc changes:
- Upgrades bountycastle version and uses newer classes
- Refactors SAMLUtil to use new CertUtils
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
If a public IP is assigned to a VPC, a VM running inside that VPC cannot ping that public IP. This is due to the IPtables Nat rules set in such a way that drop any requests to the public IP from internal interfaces. I am fixing this so that internal hosts can also reach the public IP.
Reproduction:
Create a VPC
Create a network inside the VPC
Allocate a public IP
Create a VM in the network
Create a port forwarding rule enabling ICMP
ping the public IP inside the VM (this will fail)
This makes sure IP address is active.
After a vRouter is recreated (e.g. reboot via CloudStack UI) and Remote Access VPN enabled, VPN won't work anymore. Here is the abbreviated output of "ipsec auto -status" while we were having the issue:
root@r-10-VM:~# ipsec auto --status
000 using kernel interface: netkey
000 interface lo/lo 127.0.0.1
000 interface lo/lo 127.0.0.1
000 interface eth0/eth0 169.254.1.45
000 interface eth0/eth0 169.254.1.45
000 %myid = (none)
After this commit, the following occurs and VPNs work:
root@r-10-VM:~# ipsec auto --status
000 using kernel interface: netkey
000 interface lo/lo 127.0.0.1
000 interface lo/lo 127.0.0.1
000 interface eth0/eth0 169.254.1.45
000 interface eth0/eth0 169.254.1.45
000 interface eth1/eth1 xxx.xxx.xxx.172
000 interface eth1/eth1 xxx.xxx.xxx.172
000 interface eth2/eth2 192.168.1.1
000 interface eth2/eth2 192.168.1.1
000 %myid = (none)
eth1 interface IP is masked, but now ipsec sees all the interfaces and VPN works.
Looks like this bug was introduced by Pull Request #1423
It added code to start ipsec (cloudstack/systemvm/patches/debian/config/opt/cloud/bin/configure.py)
if vpnconfig['create']:
logging.debug("Enabling remote access vpn on "+ public_ip)
CsHelper.start_if_stopped("ipsec")
This change removes an unnecessary conversion from IPNetwork
to list in one of the router scripts. This makes the router
faster at processing static NAT rules, which can prevent
timeouts when attaching or detaching IPs.
(cherry picked from commit d5c5eb10f8a4fb80aa8b9182290948755cdd6026)
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
When enabling remote access VPN, a new interface is created upon client connecting via VPN. The DNS service (dnsmasq) is set only to listen on interfaces that are active when it starts. Thus VPN users are provided the VR's IP address for DNS resolution, but it is not actually listening for DNS requests.
Mozilla Firefox displays white tile in place of cursor. The reason - function isImageLoaded() always returns true after first load and function checkUpdate() reloads too fast.
Suggested fix - in refresh() method state imageLoaded should be reverted to false. This ensures that function checkUpdate() processes only when tile image is loaded.
Current SSL protocol and ciphers used in SystemVMs are not the
recommended. To analyze it is possible to use tests such as from SSL
Labs (https://www.ssllabs.com/ssltest/). This commit changes the grade
from C to -A
This enables the firewall/mangle tables rules to ACCEPT instead of RETURN, which
is the same behaviour as observed in ACS 4.5. By accepting the traffic, guest
VMs will be able to communicate tcp traffic between each other over snat public
IPs.
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
