I have a repeatable failure for SoftLayer accounts with large numbers of instances.
require 'rubygems'
require 'softlayer_api'
require 'pp'
client = SoftLayer::Client.new(
:username => ENV["SL_API_USER"], ## API Username
:api_key => ENV["SL_API_KEY"], ## API Key
:timeout => 120
)
BILLING_ITEM = "mask[
id,
description,
categoryCode,
category,
createDate,
cancellationDate,
lastBillDate,
nextBillDate,
orderItemId,
parentId,
recurringMonths,
serviceProviderId,
invoiceItem[hostName,domainName,resourceTableId,notes],
item,
location[longName],
nextInvoiceTotalOneTimeAmount,
nextInvoiceTotalOneTimeTaxAmount,
nextInvoiceTotalRecurringAmount,
nextInvoiceTotalRecurringTaxAmount,
bundledItems,
currentHourlyCharge,
hourlyRecurringFee,
hoursUsed,
orderItem[order[userRecord]]
]"
acct = client.service_named("SoftLayer_Account")
items = acct.object_mask(BILLING_ITEM).getNextInvoiceTopLevelBillingItems
unless items.kind_of?(Array)
items = [items]
end
pp items.empty? ? "No results" : items.first
On small accounts (say like 50 devices), this will return just fine. But on large accounts (1000+ devices), this will fail with a HTTP 500 error after about 40 seconds (well before the 120 second timeout). Bringing the timeout down to low values does cause it to abort with a predictable timeout error, so it seems like there's something happening on the SL side.
Other queries like fetching virtual guests with deep object masks will respect the long timeouts past 40 seconds, so this seems like it may be unique to the billing layer you have gating this data.
...unless of course I'm missing something obvious. :)
I have a repeatable failure for SoftLayer accounts with large numbers of instances.
On small accounts (say like 50 devices), this will return just fine. But on large accounts (1000+ devices), this will fail with a HTTP 500 error after about 40 seconds (well before the 120 second timeout). Bringing the timeout down to low values does cause it to abort with a predictable timeout error, so it seems like there's something happening on the SL side.
Other queries like fetching virtual guests with deep object masks will respect the long timeouts past 40 seconds, so this seems like it may be unique to the billing layer you have gating this data.
...unless of course I'm missing something obvious. :)